Current IssueSpecial IssueEditorial BoardOnline First ArticlesArchive
Current IssueSpecial IssueEditorial BoardOnline First ArticlesArchive

Full Research Article

An Analytical Study on Indigenous and Crossbred Cow Breeds of Jammu and Kashmir and Ladakh

A
Arnav Saxena1
N
Naveed Tariq1
S
Syed Rameem Zahra1,*
M
Mohammed Hameed Alhameed2
1Centre for Artificial Intelligence and Machine Learning, Sher-e-Kashmir University of Agricultural Sciences and Technology of Kashmir, Shalimar, Srinagar-190 025, Jammu and Kashmir, India.
2College of Engineering and Computer Science, Jazan University, KSA.

ABSTRACT

Background: Cattle are of great importance in terms of economy, food and livelihood of Jammu and Kashmir, as well as Ladakh, owing to the varied agro-climatic conditions that favor the existence of varied cow breeds of indigenous, crossbred and even exotic types. The cattle breeding practices, which were followed in a traditional manner, have been an important part of the local environment and are largely dependent on local adaptation and local wisdom that have been accumulated over the years. The cattle are adversely affected by breed identification problems, overcrossbreeding and various infectious and parasitic diseases.

Methods: Taking into account all these factors, this research aims at providing an analytical overview of cow breeds, their population, as well as diseases faced, all of this concerning the state of Jammu and Kashmir, as well as Ladakh. An experimental analysis has been conducted on how artificial intelligence can be made use of for identifying cow breeds, where a computer vision-related model has been developed that utilizes the combined power of a pre-trained model of YOLOv8x for efficient extraction of cow-related regions, followed by the use of an EfficientNetV2-S model for the identification of various cow breeds on a dataset of Indian cow images.

Result: The model achieved an accuracy of 75.79%, a Top-5 accuracy of 98.11% and a Macro F1-score of approximately 0.78, indicating consistent performance across varying conditions. The results suggest that the model has potential for use in cattle breed identification as a supportive tool alongside traditional methods. The integration of conventional and AI-based approaches may contribute to improved livestock management practices.

INTRODUCTION

Cattle rearing plays an important role in the agricultural economy and rural livelihood systems of Jammu and Kashmir and Ladakh. The diverse agro-climatic conditions of these regions support a variety of indigenous, crossbred and exotic cattle breeds that contribute significantly to milk production and household income. Indigenous cattle are generally well adapted to harsh environmental conditions and possess better disease resistance, whereas crossbreeding with exotic breeds such as jersey and holstein friesian (HF) has been widely adopted to improve milk productivity.
       
Accurate cattle breed identification is essential for livestock management, conservation of indigenous genetic resources and implementation of region-specific breeding strategies. However, traditional breed identification methods based on visual observation are often subjective and become challenging in regions with extensive crossbreeding. These limitations can result in misidentification and affect breeding, disease management and conservation practices.
       
Recent advancements in artificial intelligence (AI), deep learning and computer vision have shown promising applications in livestock monitoring, animal recognition and automated classification systems (Andrew et al., 2019; Silva et al., 2021; Chen et al., 2024). Deep learning-based image analysis methods have demonstrated effective performance in agricultural and livestock-related classification tasks under real-world conditions. Such approaches provide opportunities for developing objective and scalable cattle breed identification systems that can support farmers and livestock managers.
       
Despite these developments, limited studies have focused on AI-assisted cattle breed identification in the Himalayan regions of Jammu and Kashmir and Ladakh, where diverse climatic conditions and crossbreeding practices make breed recognition more difficult. Therefore, the present study aims to provide an analytical overview of major cattle breeds, population distribution and prevalent diseases affecting cattle in these regions while also experimentally evaluating an AI-based cattle breed classification framework using YOLOv8x and EfficientNetV2-S models. The study highlights the potential of integrating AI-assisted approaches with traditional livestock management practices for improved cattle monitoring and conservation.
 
Cow breeds in Jammu, Kashmir and Ladakh
 
Jammu and Kashmir, as well as Ladakh, are home to a variety of indigenous, cross-bred, as well as exotic breeds of cows because of the varying climate, altitude and rearing systems. Generally, the indigenous breeds of cows are found to be well adapted to the existing climatic conditions. On the other hand, cross-bred and exotic breeds of cows are found in some areas to improve the milk yield of those areas. The varying breeds of cows are distributed unevenly in these three regions (Rather et al., 2022; ICAR, 2019; ICAR-NBAGR, 2020; Lawrence, 2002; Rege and Okeyo, 2006).
 
Cow breeds in Kashmir
 
The climate of the Kashmir Valley is generally temperate, with cold winters and mild summers. The indigenous cow of the Kashmir Valley is known as the Kashmiri cow. The cow is medium in size, sturdy and well adapted to the local conditions of low temperature and limited availability of feed. Among the exotic breeds, Jersey cows are commonly used in the Kashmir Valley due to their lower feed requirements, higher milk fat content and better adaptability compared to larger exotic breeds. HF cattle have also been found, mostly in organized as well as semi-organized dairy enterprises. Crossbred cows, which are a combination of local cows and Jersey or HF, are also prevalent in this valley (Hamadani, 2013; Iqbal and Pampori, 2008; Makhdoomi et al., 2013).
 
Cow breeds in Jammu
 
The climate in the Jammu area is sub-tropical to semi-arid with a higher temperature in summer than in Kashmir. Indigenous breeds like hariana and tharparkar are prevalent in the Jammu area. These breeds are known for their strength, endurance, tolerance to heat and feed scarcity and have been utilized as dual-purpose breeds for milk and draft purposes. Well-known Indian dairy breeds like Gir and Red Sindhi cows are also found in some of the dairy farms. Crossbred cows are prevalent in a large number of cattle, mainly a combination of local breeds and Jersey, HF, or Gir (Gupta et al., 1996; Shergojry et al., 2017).
 
Cow breeds in Ladakh
 
Ladakh represents cold arid high-altitude areas with harsh climatic conditions, low oxygen and scarce fodder. The Ladakhi cow is also known as the local mountain cow. This cow is of low stature and is found in high-altitude and low-temperature areas. Apart from the local cows, there are also hybrids such as dzomo (female) and Dzo (male), which are crossbreeds of yak and cattle. In the animal economy of Ladakh, the cow variety of dzomo produces more milk than the yak and is of immense importance (ICAR-NBAGR, 2020; Gupta et al., 1996; Makhdoomi et al., 2013; Kumari et al., 2019; Rege and Okeyo, 2006).
 
Population status of cows
 
The cattle population in Jammu and Kashmir and Ladakh represents the varied agro-climatic conditions, land use and livestock development in the region (DAHD, 2019a; DAHD, 2019b). Cow population is one of the largest constituents of the livestock population and holds the key to dairy development, particularly for small and marginal farmers. The density of cattle population reveals marked variation in the different districts because of several factors such as availability of grazing land, veterinary facilities, market accessibility and use of crossbreeding methods (Table 1) (FAO, 2018; FAO, 2017; DAHD, 2019a; DAHD, 2019b; Office of the Registrar General, 2021; DAHD, 2019c; FAO, 2020).

Table 1: District-wise cow population in Jammu and Kashmir and Ladakh with major breeds and remarks. Source: Compiled from DAHD (2019a; 2019b; 2019c), office of the registrar general (2021), FAO (2020).


 
Major diseases affecting cows
 
Diseases are among the major constraints that limit the productivity and breeding of cattle in Jammu and Kashmir and Ladakh. The prevalence and effect of diseases are often influenced by regional climatic factors, management systems and mobility of animals, together with levels of veterinary support. Indigenous cows as well as crossbred cows are often affected by diseases; however, some impacts are noted as varying across breeds (Table 2) (WOAH, 2021a; WOAH, 2021b; Rahman et al., 2023; Arora et al., 2019; Singh et al., 2018; Sharma et al., 2020). Recent advancements in artificial intelligence have also enabled automated disease detection in cattle using image-based approaches, highlighting the broader potential of AI in livestock health monitoring and management (AlZubi, 2024).

Table 2: Major diseases affecting cows in Jammu and Kashmir and Ladakh. Source: Compiled from Rahman et al. (2023); Arora et al. (2019); Singh et al. (2018); Sharma et al. (2020); FAO (2016).


 
Viral diseases
 
Foot and mouth disease (FMD) is one of the most widespread and economically important viral diseases affecting cows across the region. Cows suffering from FMD display clinical syndromes like fever, salivation, mouth and foot lesions, lameness and a sharp decrease in milk production (Akhoon et al., 2015; Dominguez et al., 2003; Govindaraj et al., 2020; Hasan and Mia, 2021; King et al., 2015). Lumpy Skin Disease (LSD) has surfaced as one of the major cattle diseases in recent times. The mode of transmission is primarily by insects such as mosquitoes and flies. Outbreaks have been noted in various districts of Jammu and Kashmir (Rahman et al., 2023; Bhattacharya et al., 2023).
 
Bacterial diseases
 
Mastitis, a bacterial infection of the udder, is a common problem in high-yielding crossbred cows like HF and jersey. Hemorrhagic septicemia (HS) and black quarter (BQ) are acute bacterial infections prevalent in indigenous and young cattle, especially in grazing-based rearing systems (Arora et al., 2019; Singh et al., 2018; Krishnamoorthy et al., 2017; Krishnamoorthy et al., 2019a; Krishnamoorthy et al., 2019b; Krishnamoorthy et al., 2020).
 
Parasitic and metabolic disorders
 
Parasitic infestations like internal parasites (roundworms, tapeworms) and external parasites (ticks, lice) are quite common in most pastures of Jammu and Kashmir and Ladakh. Metabolic disorders such as milk fever (hypocalcemia) and bloat are more prevalent in high-yielding dairy cows (Krishnamoorthy et al., 2017; Krishnamoorthy et al., 2020; Sharma, 2021; Malik and Sharma, 2019).
       
The epidemiology of cattle diseases in the region is inescapably linked to the existing climatic factors and cattle movement. The seasonal migration of cattle to the high-altitude grazing grounds in the summer season, the housing of cattle in large groups during the winter season and the process of crossbreed grazing have increased the chances of disease transmission. The crossbred cows, although more productive, are more prone to diseases and result in higher economic losses (Sharma et al., 2020; Krishnamoorthy et al., 2019a; Krishnamoorthy et al., 2019b; Krishnamoorthy et al., 2021a; Krishnamoorthy et al., 2021b).

MATERIALS AND METHODS

Experimental design and objective
 
In this study, the experimental research design methodology has been used to evaluate the feasibility of using AI-based computer vision and machine learning (ML) tools for the automatic identification of cattle breeds in unconstrained environments. The primary objective was to develop a robust system with a two-stage deep learning model that can automatically identify Indian cattle breeds from images. This research study does not aim to comprehensively review all the breeds of cows found in Jammu and Kashmir and Ladakh. The experiment was performed to automatically identify cattle breeds from images using ML algorithms.
 
Dataset source and preparation
 
The experiments were performed on the Indian Bovine Breeds dataset (Kaggle Community, 2023), which is available on Kaggle. The original dataset included images captured in real-world conditions on farms. To ensure data consistency and relevance, manual data cleaning was performed before model training. All buffalo images were eliminated and only cattle images are taken into account. After the dataset cleaning process, there are 16 Indian breeds of cattle available for experimentation (Table 3). The dataset includes 2,176 training images, 625 validation images and 318 test images.

Table 3: Final cattle breed distribution after dataset cleaning.


 
Cattle detection and region extraction
 
To reduce background noises in the data further and for proper classification, an extra step of preprocessing was introduced to include object detection. A YOLOv8x model, which is pre-trained on common objects in context (COCO) data, was used for cattle image detection. The model was not trained or modified further; instead, the COCO cow class was directly utilized for cattle image detection. Regions of cattle that were detected were cropped to obtain images of the chosen region of interest (Fig 1).

Fig 1: Workflow of cattle detection and image cropping using YOLOv8x before breed classification.


 
Breed classification using EfficientNetV2
 
For cattle breed classification, the EfficientNetV2-S CNN architecture was employed (He et al., 2016; Tan and Le, 2021). The detected images that passed through the detection phase were then resized to a fixed resolution of 256 × 256 pixels, which was then fed as input to the classifier. Transfer learning was implemented by fine-tuning the existing model to adapt to cattle breed classification. The model was trained using a supervised learning algorithm with a sparse categorical cross-entropy loss function and an Adam optimizer.
 
Evaluation metrics
 
The performance assessment of the model required the application of a variety of metrics. The metrics included the Top-1 accuracy, the Top-5 accuracy to ensure the performance of the model under visually similar breeds and the Macro F1-score to ensure performance on all classes, particularly under class imbalance settings.

RESULTS AND DISCUSSION

Overall model performance
 
The performance analysis of the proposed AI-based cattle breed recognition system was conducted using a test dataset consisting of 318 images. The proposed system demonstrated satisfactory performance in classifying different cattle breeds under field conditions. The pre-trained EfficientNetV2-S model with YOLOv8x-preprocessed crops achieved a Top-1 classification accuracy of approximately 75.8%, a Top-5 accuracy of 98.1% and a Macro F1-score of approximately 0.78 (Table 4).

Table 4: Overall and breed-wise classification performance metrics of the proposed model (318 test samples).


       
To contextualize the performance of the proposed model, a comparison with existing studies on cattle breed classification was conducted. Previous research using deep learning approaches such as CNNs, YOLO-based detectors and hybrid pipelines has reported varying levels of accuracy depending on dataset characteristics and experimental conditions. For instance, YOLOv4-based detection models have achieved approximately 81% accuracy on multi-breed datasets, while ensemble CNN approaches have reported F1-scores around 0.86 in limited breed settings. Similarly, ResNet-based models trained on curated datasets have demonstrated accuracy as high as 94%, whereas studies involving real-world or morphologically similar breeds report moderate accuracy in the range of 70-85%.
       
In comparison, the proposed model achieved a Top-1 accuracy of 75.8% and a Top-5 accuracy of 98.1% under unconstrained conditions with multiple breeds. These results indicate that the model provides competitive performance, particularly considering the challenges of real-world datasets, class imbalance and inter-breed similarity (Table 5).

Table 5: Comparison of proposed model with existing studies.


       
Breed-wise classification accuracy was found to vary with different breeds. Morphologically distinct breeds such as Banni, HF, Sahiwal, Toda and Brown Swiss were found to have high classification accuracy. Similarly, breeds with visual similarity such as Ongole, Deoni, Jersey and Bargur were found to have low accuracy (Fig 2).

Fig 2: Normalized confusion matrix for cattle breed classification.


       
Most of the misclassifications were found to be among breeds that were morphologically similar to each other. Common misclassifications occurred among breeds such as Hallikar and Bargur, Ongole and Deoni and Jersey and Brown Swiss. In addition, some breeds such as HF, Banni, Toda and Sahiwal were found to have a higher level of consistency in their classification.
       
Qualitative analysis of the accuracy of the prediction made by the model was performed by analyzing a random sample of correctly and incorrectly classified images in the test dataset (Fig 3). The model performed well on images where the body was visible. The incorrectly classified images are based on partial body, occlusion and similarity to other breeds.

Fig 3: Visualization of correct and incorrect cattle breed predictions.


       
The present study suggests the feasibility of applying AI-assisted machine learning approaches for cattle breed classification in uncontrolled field conditions. The integration of the EfficientNetV2-S approach with YOLOv8x for cattle detection and breed classification has resulted in the demonstration of a two-stage approach that demonstrates good performance. The accuracy in Top-5 results indicates that although it sometimes becomes difficult to give a correct breed identification, the correct breed is also included in the top results. This becomes an important aspect in situations that require real-world livestock management, as an AI-based approach is intended to be used as an assistance tool rather than a replacement for human intelligence.
       
From a practical perspective, such AI-based systems can be integrated into mobile or edge-based applications to assist farmers and veterinary practitioners in real-time breed identification. This can support decision-making related to breeding strategies, disease management and livestock monitoring, particularly in remote and resource-constrained regions. The use of AI tools in this context can complement traditional knowledge systems rather than replace them.
       
In the context of temperate Himalayan regions such as Jammu and Kashmir and Ladakh, AI and machine learning technologies have been identified as promising tools for improving agricultural sustainability, livestock management and resource efficiency, although challenges such as data scarcity and infrastructure limitations remain (Saxena et al., 2023).
       
From the livestock management point of view, cattle breed identification is very important for efficient cattle breeding, preservation of native breeds and cattle disease management. In traditional models, as practiced in Jammu and Kashmir and Ladakh, cattle breed identification is done mostly through visual observation and experience, which is a subjective process. The proposed AI-based approach could, with the help of technology, enhance identification models based on traditional knowledge. However, the experimental outcome has been done on a limited set of cattle breeds as per the available data sets and it is not a comprehensive test of all cow breeds in the region (Gupta et al., 1996; Shergojry et al., 2017; Krishnamoorthy et al., 2019a; Krishnamoorthy et al., 2021a; Rege and Okeyo, 2006).
       
Although this study has shown promising results, it also has some limitations. Firstly, the data was imbalanced and did not have standardized images of side-view full-body images. Furthermore, since the performance of the generic object detection model was limited under the testing conditions. Furthermore, the performance of the generic object detection model highlights the need for developing region-specific AI models trained on indigenous cattle datasets, which could significantly improve classification accuracy and real-world applicability.

CONCLUSION

This study presented an analytical overview of indigenous and crossbred cattle breeds in Jammu and Kashmir and Ladakh, including their population distribution and major disease challenges. An AI-based cattle breed classification framework using YOLOv8x and EfficientNetV2-S was also experimentally evaluated for automated breed identification under real-world conditions. The proposed model demonstrated promising classification performance, achieving satisfactory Top-1 and Top-5 accuracy despite dataset limitations and class imbalance. The findings suggest that AI-assisted approaches can support livestock management, breed documentation and conservation of indigenous cattle genetic resources. The study further highlights the importance of integrating traditional livestock practices with modern AI technologies for improving cattle monitoring and sustainable livestock development in Himalayan regions.

ACKNOWLEDGEMENT

The authors acknowledge the support of Centre of Artificial Intelligence and Machine Learning (CAIML), SKUAST-Kashmir, India.

CONFLICT OF INTEREST

The authors declare that they have no conflict of interest regarding the publication of this paper.

REFERENCES


  1. Akhoon, Z.A., Peer, F.U. and Akhoon, A.B. (2015). Incidence of foot and mouth disease at polyclinics of SKUAST-Kashmir. Veterinary Science Research Journal. 6(2): 1-6.

  2. AlZubi, A.A. (2024). Detection of disease in calves using artificial intelligence. Indian Journal of Animal Research. 58(12): 2138-2145. doi: 10.18805/IJAR.BF-1760.

  3. Andrew, M., Fry, S.A. and Trotter, M.J. (2019). Automated detection and identification of individual cattle using deep learning and YOLOv2. Computers and Electronics in Agriculture. 165: 104959.

  4. Arora, R., Singh, V. and Sharma, R.K. (2019). Incidence and economic losses due to mastitis in dairy cattle in India: A review. Indian Journal of Animal Sciences. 89(9): 939- 948.

  5. Bhattacharya, S., Singh, R.K., Barman, N.N., Kumar, A., Chopra, M., Baitha, R., Upadhyay, C., Dubey, P.K. and Singh, K.P. (2023). Lumpy skin disease in India: Epidemiology and control strategies. Veterinary World. 16(1): 50-59.

  6. Chen, X., Li, Y. and Liu, H. (2024). Behaviour-YOLO: Cattle Behavior Recognition Based on Improved YOLO Network. In: Proceedings of SPIE. 13709: 1370916.

  7. DAHD (2019a). 20th Livestock Census of India-All India Report. Ministry of Fisheries, Animal Husbandry and Dairying, Government of India, New Delhi.

  8. DAHD (2019b). Key Results of 20th Livestock Census-2019. Government of India, New Delhi.

  9. DAHD (2019c). Basic Animal Husbandry Statistics. Government of India, New Delhi.

  10. Dominguez, E., Escarmis, C., Baranowski, E. et al. (2003). Evolution of foot-and-mouth disease virus. Virus Research. 91: 47-63.

  11. El-Henawy, M., Abdel-Aziz, H. and Abdel-Wahab, M. (2017). Cattle identification using convolutional neural networks and texture features. IET Computer Vision. 11(8): 667-675.

  12. FAO (2016). Economic Analysis of Animal Diseases. FAO Animal Production and Health Guidelines. Rome.

  13. FAO (2017). The Future of Livestock in India. FAO Animal Production and Health Paper. Rome.

  14. FAO (2018). World Livestock: Transforming the Livestock Sector Through the Sustainable Development Goals. Rome.

  15. FAO (2020). Buffalo Production and Health in South Asia. FAO Regional Office for Asia and the Pacific, Bangkok.

  16. FAO (2022). The State of Animal Genetic Resources in India. Rome.

  17. Govindaraj, G., Kumar, G.B., Krishnamohan, A. et al. (2020). Economic impact of foot-and-mouth disease in cattle and buffaloes in India. Indian Journal of Animal Sciences. 90(7): 971-976.

  18. Gupta, H., Jindal, P., Verma, O.P., Arya, R.K., Ateya, A.A., Soliman, N.F. and Mohan, V. (2022). Computer vision-based approach for automatic detection of dairy cow breed. Electronics. 11(22): 3791. https://doi.org/10.3390/ electronics11223791.

  19. Gupta, S.C., Tundup, T., Gupta, N. and Kumar, P. (1996). Livestock wealth of the Ladakh: A cold arid region in India. Animal Genetic Resources. 19: 27-36.

  20. Hamadani, H. (2013). Morphometry, Production Characteristics and Rearing Practices of Domestic Geese of Kashmir. M.V.Sc. Thesis. SKUAST-Kashmir, India.

  21. Hasan, M. and Mia, M.M. (2021). Prevalence and epidemiological risk factors of foot-and-mouth disease in cattle. Bulgarian Journal of Veterinary Medicine. 24: 1-14.

  22. He, K., Zhang, X., Ren, S. and Sun, J. (2016). Deep Residual Learning for Image Recognition. In: Proceedings of the IEEE CVPR. pp. 770-778.

  23. ICAR (2019). Indigenous Cattle and Their Conservation in India. Indian Council of Agricultural Research, New Delhi.

  24. ICAR-NBAGR (2020). Breed Descriptor of Ladakhi Cattle. National Bureau of Animal Genetic Resources, Karnal, India.

  25. ICAR-NIVEDI (2023). Animal Disease Informatics Report: Jammu and Kashmir. National Institute of Veterinary Epidemiology and Disease Informatics, Bengaluru, India.

  26. Iqbal, S. and Pampori, Z.A. (2008). Production potential and qualitative traits of indigenous chicken of Kashmir. Livestock Research for Rural Development. 20(11): 14.

  27. Kaggle Community (2023). Indian Bovine Breeds Dataset. Kaggle. https://www.kaggle.com/datasets/indian-bovine-breeds.

  28. Kappali, H., Sadyojatha, K.M. and Prashanthi, S.K. (2023). Computer vision and machine learning in paddy diseases identification and classification: A review. Indian Journal of Agricultural Research. 57(2): 183-187. doi: 10.18805/IJARe.A-6061.

  29. King, D.P., Ludi, A., Wilsden, G., Parida, S. and Paton, D.J. (2015). Differentiation of Vaccinated and Infected Animals Using Non-Structural Proteins. In: OIE Regional Commission Proceedings. pp. 1-4.

  30. Krishnamoorthy, P., Akshata, L.G., Jacob, S.S. et al. (2021b). Prevalence of theileriosis in cattle and buffaloes in India. Indian Journal of Animal Sciences. 91(4): 269-279.

  31. Krishnamoorthy, P., Ashwini, M., Suresh, K.P. and Roy, P. (2019a). Prevalence of Anaplasma species in dairy animals: A systematic review. Research in Veterinary Science. 123: 159-170.

  32. Krishnamoorthy, P., Goudar, A.L., Suresh, K.P. and Roy, P. (2021a). Global prevalence of mastitis in dairy cattle and buffaloes.  Research in Veterinary Science. 136: 561-586.

  33. Krishnamoorthy, P., Hamsapriya, S., Ashwini, M. et al. (2019b). Livestock-associated methicillin-resistant Staphylococcus aureus  in India. International Journal of Livestock Research. 9: 179-191.

  34. Krishnamoorthy, P., Suresh, K.P. and Roy, P. (2020). Meta-analysis as a tool for estimating prevalence of livestock diseases. Research and Reviews: Journal of Veterinary Sciences. 8: 1-10.

  35. Krishnamoorthy, P., Suresh, K.P., Saha, S. et al. (2017). Meta- analysis of prevalence of mastitis in dairy cattle in India. International Journal of Current Microbiology and Applied Sciences. 6: 1214-1234.

  36. Kumari, P., Bharti, V.K., Kumar, D., Mukesh, M. and Sharma, I. (2019). Haematological and biochemical profiling of Ladakhi cow. Indian Journal of Animal Sciences. 90(4): 527-531.

  37. Lawrence, W.R. (2002). The Valley of Kashmir. Gulshan Publishers, Srinagar, India.

  38. Li, Z., Chen, H., Zhang, Y. et al. (2022). A lightweight deep learning model for cattle face recognition. Computers and Electronics in Agriculture. 200: 107208.

  39. Mahato, S., Kumar, R., Choudhary, A. et al. (2025). Dairy DigiD: A keypoint-based deep learning system for individual dairy cow identification. Frontiers in Artificial Intelligence. 8: 1545247.

  40. Makhdoomi, D., Gazi, M., Nabi, S. and Ahmed, S. (2013). Morphometric studies on adult double-humped camels of Ladakh, India. Emirates Journal of Food and Agriculture. 25(7): 544- 548.

  41. Malik, R. and Sharma, R. (2019). Tick-borne diseases in buffaloes of northern India. Veterinary World. 12(6): 894-900.

  42. Mamatha, D.M., Venkata, L.P., Thanikonda, D.N., Kotagaram, H. and Koduru, S.K. (2026). Integrated DNA barcoding and QR code-based image recognition for identification of indian cattle breeds. Indian Journal of Animal Research. doi: 10.18805/IJAR.B-5745.

  43. Office of the Registrar General (2021). District-Wise Details of Livestock Population in Jammu and Kashmir-20th Livestock Census 2019. Open Government Data Platform India.

  44. Pawaskar, P., Yogish, H.K., Pakruddin, B. and Deepa, Y. (2025). Novel and efficient deep learning models for assessing AI’s impact on disease diagnosis in agriculture. Indian Journal of Agricultural Research. 59(9): 1395-1403. doi: 10.18805/IJARe.A-6388.

  45. Rahman, S.M., Singh, P.K. and Tiwari, A.K. (2023). Epidemiology and economic impact of lumpy skin disease in cattle in India. Tropical Animal Health and Production. 55: 219.

  46. Rather, A.F.K., Ganai, N.A., Shah, A.A. and Sheikh, A.A. (2022). Livestock and poultry breeds of Jammu and Kashmir and Ladakh. Indian Journal of Animal Sciences. 92(4): 409-416.

  47. Rather, A.F.K., Sheikh, A.A., Shah, A.A. et al. (2025). Computer vision in precision livestock farming. Computers and Electronics in Agriculture. 222: 110555.

  48. Rege, J.E.O. and Okeyo, M.A. (2006). Improving Our Knowledge of Tropical Indigenous Animal Genetic Resources. FAO/ AGTR Training Module Version 2. Rome.

  49. Saxena, A., Faiq, M., Ghatrehsamani, S. and Zahra, S.R. (2023). integrating artificial intelligence and machine learning for sustainable development in agriculture and allied sectors of the temperate himalayas. Agri. Engineering. 8(1): 35.

  50. Sharma, R.K. (2021). Fasciolosis in livestock of the Himalayan region. Journal of Parasitic Diseases. 45(3): 678-684.

  51. Sharma, S., Katoch, R. and Verma, N. (2020). Prevalence of major bovine diseases in Jammu and Kashmir: A retrospective study. Journal of Entomology and Zoology Studies. 8(3): 1230-1235.

  52. Shergojry, S.A., Akhoon, Z.A., Mubarak, T. and Namgyal, D. (2017). Socio-economic impact of livestock in tribal areas of Leh. Journal of Krishi Vigyan. 6(1): 187-190.

  53. Silva, J.E.P., Krohling, R.A. and Henriques, R.P.B. (2021). Computer vision-based systems for individual cattle identification: A review. Computers and Electronics in Agriculture. 185: 106131.

  54. Singh, A.K., Gupta, P.K. and Kumar, S. (2018). Prevalence and economic losses due to brucellosis in dairy animals in India. Indian Journal of Animal Sciences. 88(10): 1173- 1180.

  55. Tan, M. and Le, Q. (2021). EfficientNetV2: Smaller Models and Faster Training. In: Proceedings of the 38th ICML. 139: 10096-10106.

  56. Vijayalakshmi, A., Shanmugavadivu, P., Vijayalakshmi, S., Padarha, S. and Sivaranjani, R. (2024). Ensemble Learning Algorithm for Cattle Breed Identification using Computer Vision Techniques. In: Proceedings of the International Conference on Artificial Intelligence, Communication, IoT, Data Engineering and Security (IACIDS 2023). EAI. https:// doi.org/10.4108/eai.23-11-2023.2343338.

  57. Warhade, R., Devi, I., Singh, N., Arya, S., Dudi, K., Lathwal, S.S. and Tomar, D.S. (2024). Attention module incorporated transfer learning empowered deep learning-based models for classification of phenotypically similar tropical cattle breeds (Bos indicus). Tropical Animal Health and Production. 56(6): 192. https://doi.org/10.1007/s11250- 024-04050-7.

  58. WOAH (2021a). Foot and Mouth Disease (Infection with Foot and Mouth Disease Virus). WOAH Technical Disease Card. Paris.

  59. WOAH (2021b). Lumpy Skin Disease. WOAH Technical Disease Card. Paris.
Published In
Indian Journal of Animal Research
Article Metrics
Metrics Icon
64
Views
0
Citations
Reviewed By
Share Article
Download Article
In this Article
    Contact us
    Follow us

    Full Research Article

    An Analytical Study on Indigenous and Crossbred Cow Breeds of Jammu and Kashmir and Ladakh

    A
    Arnav Saxena1
    N
    Naveed Tariq1
    S
    Syed Rameem Zahra1,*
    M
    Mohammed Hameed Alhameed2
    1Centre for Artificial Intelligence and Machine Learning, Sher-e-Kashmir University of Agricultural Sciences and Technology of Kashmir, Shalimar, Srinagar-190 025, Jammu and Kashmir, India.
    2College of Engineering and Computer Science, Jazan University, KSA.

    ABSTRACT

    Background: Cattle are of great importance in terms of economy, food and livelihood of Jammu and Kashmir, as well as Ladakh, owing to the varied agro-climatic conditions that favor the existence of varied cow breeds of indigenous, crossbred and even exotic types. The cattle breeding practices, which were followed in a traditional manner, have been an important part of the local environment and are largely dependent on local adaptation and local wisdom that have been accumulated over the years. The cattle are adversely affected by breed identification problems, overcrossbreeding and various infectious and parasitic diseases.

    Methods: Taking into account all these factors, this research aims at providing an analytical overview of cow breeds, their population, as well as diseases faced, all of this concerning the state of Jammu and Kashmir, as well as Ladakh. An experimental analysis has been conducted on how artificial intelligence can be made use of for identifying cow breeds, where a computer vision-related model has been developed that utilizes the combined power of a pre-trained model of YOLOv8x for efficient extraction of cow-related regions, followed by the use of an EfficientNetV2-S model for the identification of various cow breeds on a dataset of Indian cow images.

    Result: The model achieved an accuracy of 75.79%, a Top-5 accuracy of 98.11% and a Macro F1-score of approximately 0.78, indicating consistent performance across varying conditions. The results suggest that the model has potential for use in cattle breed identification as a supportive tool alongside traditional methods. The integration of conventional and AI-based approaches may contribute to improved livestock management practices.

    INTRODUCTION

    Cattle rearing plays an important role in the agricultural economy and rural livelihood systems of Jammu and Kashmir and Ladakh. The diverse agro-climatic conditions of these regions support a variety of indigenous, crossbred and exotic cattle breeds that contribute significantly to milk production and household income. Indigenous cattle are generally well adapted to harsh environmental conditions and possess better disease resistance, whereas crossbreeding with exotic breeds such as jersey and holstein friesian (HF) has been widely adopted to improve milk productivity.
           
    Accurate cattle breed identification is essential for livestock management, conservation of indigenous genetic resources and implementation of region-specific breeding strategies. However, traditional breed identification methods based on visual observation are often subjective and become challenging in regions with extensive crossbreeding. These limitations can result in misidentification and affect breeding, disease management and conservation practices.
           
    Recent advancements in artificial intelligence (AI), deep learning and computer vision have shown promising applications in livestock monitoring, animal recognition and automated classification systems (Andrew et al., 2019; Silva et al., 2021; Chen et al., 2024). Deep learning-based image analysis methods have demonstrated effective performance in agricultural and livestock-related classification tasks under real-world conditions. Such approaches provide opportunities for developing objective and scalable cattle breed identification systems that can support farmers and livestock managers.
           
    Despite these developments, limited studies have focused on AI-assisted cattle breed identification in the Himalayan regions of Jammu and Kashmir and Ladakh, where diverse climatic conditions and crossbreeding practices make breed recognition more difficult. Therefore, the present study aims to provide an analytical overview of major cattle breeds, population distribution and prevalent diseases affecting cattle in these regions while also experimentally evaluating an AI-based cattle breed classification framework using YOLOv8x and EfficientNetV2-S models. The study highlights the potential of integrating AI-assisted approaches with traditional livestock management practices for improved cattle monitoring and conservation.
     
    Cow breeds in Jammu, Kashmir and Ladakh
     
    Jammu and Kashmir, as well as Ladakh, are home to a variety of indigenous, cross-bred, as well as exotic breeds of cows because of the varying climate, altitude and rearing systems. Generally, the indigenous breeds of cows are found to be well adapted to the existing climatic conditions. On the other hand, cross-bred and exotic breeds of cows are found in some areas to improve the milk yield of those areas. The varying breeds of cows are distributed unevenly in these three regions (Rather et al., 2022; ICAR, 2019; ICAR-NBAGR, 2020; Lawrence, 2002; Rege and Okeyo, 2006).
     
    Cow breeds in Kashmir
     
    The climate of the Kashmir Valley is generally temperate, with cold winters and mild summers. The indigenous cow of the Kashmir Valley is known as the Kashmiri cow. The cow is medium in size, sturdy and well adapted to the local conditions of low temperature and limited availability of feed. Among the exotic breeds, Jersey cows are commonly used in the Kashmir Valley due to their lower feed requirements, higher milk fat content and better adaptability compared to larger exotic breeds. HF cattle have also been found, mostly in organized as well as semi-organized dairy enterprises. Crossbred cows, which are a combination of local cows and Jersey or HF, are also prevalent in this valley (Hamadani, 2013; Iqbal and Pampori, 2008; Makhdoomi et al., 2013).
     
    Cow breeds in Jammu
     
    The climate in the Jammu area is sub-tropical to semi-arid with a higher temperature in summer than in Kashmir. Indigenous breeds like hariana and tharparkar are prevalent in the Jammu area. These breeds are known for their strength, endurance, tolerance to heat and feed scarcity and have been utilized as dual-purpose breeds for milk and draft purposes. Well-known Indian dairy breeds like Gir and Red Sindhi cows are also found in some of the dairy farms. Crossbred cows are prevalent in a large number of cattle, mainly a combination of local breeds and Jersey, HF, or Gir (Gupta et al., 1996; Shergojry et al., 2017).
     
    Cow breeds in Ladakh
     
    Ladakh represents cold arid high-altitude areas with harsh climatic conditions, low oxygen and scarce fodder. The Ladakhi cow is also known as the local mountain cow. This cow is of low stature and is found in high-altitude and low-temperature areas. Apart from the local cows, there are also hybrids such as dzomo (female) and Dzo (male), which are crossbreeds of yak and cattle. In the animal economy of Ladakh, the cow variety of dzomo produces more milk than the yak and is of immense importance (ICAR-NBAGR, 2020; Gupta et al., 1996; Makhdoomi et al., 2013; Kumari et al., 2019; Rege and Okeyo, 2006).
     
    Population status of cows
     
    The cattle population in Jammu and Kashmir and Ladakh represents the varied agro-climatic conditions, land use and livestock development in the region (DAHD, 2019a; DAHD, 2019b). Cow population is one of the largest constituents of the livestock population and holds the key to dairy development, particularly for small and marginal farmers. The density of cattle population reveals marked variation in the different districts because of several factors such as availability of grazing land, veterinary facilities, market accessibility and use of crossbreeding methods (Table 1) (FAO, 2018; FAO, 2017; DAHD, 2019a; DAHD, 2019b; Office of the Registrar General, 2021; DAHD, 2019c; FAO, 2020).

    Table 1: District-wise cow population in Jammu and Kashmir and Ladakh with major breeds and remarks. Source: Compiled from DAHD (2019a; 2019b; 2019c), office of the registrar general (2021), FAO (2020).


     
    Major diseases affecting cows
     
    Diseases are among the major constraints that limit the productivity and breeding of cattle in Jammu and Kashmir and Ladakh. The prevalence and effect of diseases are often influenced by regional climatic factors, management systems and mobility of animals, together with levels of veterinary support. Indigenous cows as well as crossbred cows are often affected by diseases; however, some impacts are noted as varying across breeds (Table 2) (WOAH, 2021a; WOAH, 2021b; Rahman et al., 2023; Arora et al., 2019; Singh et al., 2018; Sharma et al., 2020). Recent advancements in artificial intelligence have also enabled automated disease detection in cattle using image-based approaches, highlighting the broader potential of AI in livestock health monitoring and management (AlZubi, 2024).

    Table 2: Major diseases affecting cows in Jammu and Kashmir and Ladakh. Source: Compiled from Rahman et al. (2023); Arora et al. (2019); Singh et al. (2018); Sharma et al. (2020); FAO (2016).


     
    Viral diseases
     
    Foot and mouth disease (FMD) is one of the most widespread and economically important viral diseases affecting cows across the region. Cows suffering from FMD display clinical syndromes like fever, salivation, mouth and foot lesions, lameness and a sharp decrease in milk production (Akhoon et al., 2015; Dominguez et al., 2003; Govindaraj et al., 2020; Hasan and Mia, 2021; King et al., 2015). Lumpy Skin Disease (LSD) has surfaced as one of the major cattle diseases in recent times. The mode of transmission is primarily by insects such as mosquitoes and flies. Outbreaks have been noted in various districts of Jammu and Kashmir (Rahman et al., 2023; Bhattacharya et al., 2023).
     
    Bacterial diseases
     
    Mastitis, a bacterial infection of the udder, is a common problem in high-yielding crossbred cows like HF and jersey. Hemorrhagic septicemia (HS) and black quarter (BQ) are acute bacterial infections prevalent in indigenous and young cattle, especially in grazing-based rearing systems (Arora et al., 2019; Singh et al., 2018; Krishnamoorthy et al., 2017; Krishnamoorthy et al., 2019a; Krishnamoorthy et al., 2019b; Krishnamoorthy et al., 2020).
     
    Parasitic and metabolic disorders
     
    Parasitic infestations like internal parasites (roundworms, tapeworms) and external parasites (ticks, lice) are quite common in most pastures of Jammu and Kashmir and Ladakh. Metabolic disorders such as milk fever (hypocalcemia) and bloat are more prevalent in high-yielding dairy cows (Krishnamoorthy et al., 2017; Krishnamoorthy et al., 2020; Sharma, 2021; Malik and Sharma, 2019).
           
    The epidemiology of cattle diseases in the region is inescapably linked to the existing climatic factors and cattle movement. The seasonal migration of cattle to the high-altitude grazing grounds in the summer season, the housing of cattle in large groups during the winter season and the process of crossbreed grazing have increased the chances of disease transmission. The crossbred cows, although more productive, are more prone to diseases and result in higher economic losses (Sharma et al., 2020; Krishnamoorthy et al., 2019a; Krishnamoorthy et al., 2019b; Krishnamoorthy et al., 2021a; Krishnamoorthy et al., 2021b).

    MATERIALS AND METHODS

    Experimental design and objective
     
    In this study, the experimental research design methodology has been used to evaluate the feasibility of using AI-based computer vision and machine learning (ML) tools for the automatic identification of cattle breeds in unconstrained environments. The primary objective was to develop a robust system with a two-stage deep learning model that can automatically identify Indian cattle breeds from images. This research study does not aim to comprehensively review all the breeds of cows found in Jammu and Kashmir and Ladakh. The experiment was performed to automatically identify cattle breeds from images using ML algorithms.
     
    Dataset source and preparation
     
    The experiments were performed on the Indian Bovine Breeds dataset (Kaggle Community, 2023), which is available on Kaggle. The original dataset included images captured in real-world conditions on farms. To ensure data consistency and relevance, manual data cleaning was performed before model training. All buffalo images were eliminated and only cattle images are taken into account. After the dataset cleaning process, there are 16 Indian breeds of cattle available for experimentation (Table 3). The dataset includes 2,176 training images, 625 validation images and 318 test images.

    Table 3: Final cattle breed distribution after dataset cleaning.


     
    Cattle detection and region extraction
     
    To reduce background noises in the data further and for proper classification, an extra step of preprocessing was introduced to include object detection. A YOLOv8x model, which is pre-trained on common objects in context (COCO) data, was used for cattle image detection. The model was not trained or modified further; instead, the COCO cow class was directly utilized for cattle image detection. Regions of cattle that were detected were cropped to obtain images of the chosen region of interest (Fig 1).

    Fig 1: Workflow of cattle detection and image cropping using YOLOv8x before breed classification.


     
    Breed classification using EfficientNetV2
     
    For cattle breed classification, the EfficientNetV2-S CNN architecture was employed (He et al., 2016; Tan and Le, 2021). The detected images that passed through the detection phase were then resized to a fixed resolution of 256 × 256 pixels, which was then fed as input to the classifier. Transfer learning was implemented by fine-tuning the existing model to adapt to cattle breed classification. The model was trained using a supervised learning algorithm with a sparse categorical cross-entropy loss function and an Adam optimizer.
     
    Evaluation metrics
     
    The performance assessment of the model required the application of a variety of metrics. The metrics included the Top-1 accuracy, the Top-5 accuracy to ensure the performance of the model under visually similar breeds and the Macro F1-score to ensure performance on all classes, particularly under class imbalance settings.

    RESULTS AND DISCUSSION

    Overall model performance
     
    The performance analysis of the proposed AI-based cattle breed recognition system was conducted using a test dataset consisting of 318 images. The proposed system demonstrated satisfactory performance in classifying different cattle breeds under field conditions. The pre-trained EfficientNetV2-S model with YOLOv8x-preprocessed crops achieved a Top-1 classification accuracy of approximately 75.8%, a Top-5 accuracy of 98.1% and a Macro F1-score of approximately 0.78 (Table 4).

    Table 4: Overall and breed-wise classification performance metrics of the proposed model (318 test samples).


           
    To contextualize the performance of the proposed model, a comparison with existing studies on cattle breed classification was conducted. Previous research using deep learning approaches such as CNNs, YOLO-based detectors and hybrid pipelines has reported varying levels of accuracy depending on dataset characteristics and experimental conditions. For instance, YOLOv4-based detection models have achieved approximately 81% accuracy on multi-breed datasets, while ensemble CNN approaches have reported F1-scores around 0.86 in limited breed settings. Similarly, ResNet-based models trained on curated datasets have demonstrated accuracy as high as 94%, whereas studies involving real-world or morphologically similar breeds report moderate accuracy in the range of 70-85%.
           
    In comparison, the proposed model achieved a Top-1 accuracy of 75.8% and a Top-5 accuracy of 98.1% under unconstrained conditions with multiple breeds. These results indicate that the model provides competitive performance, particularly considering the challenges of real-world datasets, class imbalance and inter-breed similarity (Table 5).

    Table 5: Comparison of proposed model with existing studies.


           
    Breed-wise classification accuracy was found to vary with different breeds. Morphologically distinct breeds such as Banni, HF, Sahiwal, Toda and Brown Swiss were found to have high classification accuracy. Similarly, breeds with visual similarity such as Ongole, Deoni, Jersey and Bargur were found to have low accuracy (Fig 2).

    Fig 2: Normalized confusion matrix for cattle breed classification.


           
    Most of the misclassifications were found to be among breeds that were morphologically similar to each other. Common misclassifications occurred among breeds such as Hallikar and Bargur, Ongole and Deoni and Jersey and Brown Swiss. In addition, some breeds such as HF, Banni, Toda and Sahiwal were found to have a higher level of consistency in their classification.
           
    Qualitative analysis of the accuracy of the prediction made by the model was performed by analyzing a random sample of correctly and incorrectly classified images in the test dataset (Fig 3). The model performed well on images where the body was visible. The incorrectly classified images are based on partial body, occlusion and similarity to other breeds.

    Fig 3: Visualization of correct and incorrect cattle breed predictions.


           
    The present study suggests the feasibility of applying AI-assisted machine learning approaches for cattle breed classification in uncontrolled field conditions. The integration of the EfficientNetV2-S approach with YOLOv8x for cattle detection and breed classification has resulted in the demonstration of a two-stage approach that demonstrates good performance. The accuracy in Top-5 results indicates that although it sometimes becomes difficult to give a correct breed identification, the correct breed is also included in the top results. This becomes an important aspect in situations that require real-world livestock management, as an AI-based approach is intended to be used as an assistance tool rather than a replacement for human intelligence.
           
    From a practical perspective, such AI-based systems can be integrated into mobile or edge-based applications to assist farmers and veterinary practitioners in real-time breed identification. This can support decision-making related to breeding strategies, disease management and livestock monitoring, particularly in remote and resource-constrained regions. The use of AI tools in this context can complement traditional knowledge systems rather than replace them.
           
    In the context of temperate Himalayan regions such as Jammu and Kashmir and Ladakh, AI and machine learning technologies have been identified as promising tools for improving agricultural sustainability, livestock management and resource efficiency, although challenges such as data scarcity and infrastructure limitations remain (Saxena et al., 2023).
           
    From the livestock management point of view, cattle breed identification is very important for efficient cattle breeding, preservation of native breeds and cattle disease management. In traditional models, as practiced in Jammu and Kashmir and Ladakh, cattle breed identification is done mostly through visual observation and experience, which is a subjective process. The proposed AI-based approach could, with the help of technology, enhance identification models based on traditional knowledge. However, the experimental outcome has been done on a limited set of cattle breeds as per the available data sets and it is not a comprehensive test of all cow breeds in the region (Gupta et al., 1996; Shergojry et al., 2017; Krishnamoorthy et al., 2019a; Krishnamoorthy et al., 2021a; Rege and Okeyo, 2006).
           
    Although this study has shown promising results, it also has some limitations. Firstly, the data was imbalanced and did not have standardized images of side-view full-body images. Furthermore, since the performance of the generic object detection model was limited under the testing conditions. Furthermore, the performance of the generic object detection model highlights the need for developing region-specific AI models trained on indigenous cattle datasets, which could significantly improve classification accuracy and real-world applicability.

    CONCLUSION

    This study presented an analytical overview of indigenous and crossbred cattle breeds in Jammu and Kashmir and Ladakh, including their population distribution and major disease challenges. An AI-based cattle breed classification framework using YOLOv8x and EfficientNetV2-S was also experimentally evaluated for automated breed identification under real-world conditions. The proposed model demonstrated promising classification performance, achieving satisfactory Top-1 and Top-5 accuracy despite dataset limitations and class imbalance. The findings suggest that AI-assisted approaches can support livestock management, breed documentation and conservation of indigenous cattle genetic resources. The study further highlights the importance of integrating traditional livestock practices with modern AI technologies for improving cattle monitoring and sustainable livestock development in Himalayan regions.

    ACKNOWLEDGEMENT

    The authors acknowledge the support of Centre of Artificial Intelligence and Machine Learning (CAIML), SKUAST-Kashmir, India.

    CONFLICT OF INTEREST

    The authors declare that they have no conflict of interest regarding the publication of this paper.

    REFERENCES


    1. Akhoon, Z.A., Peer, F.U. and Akhoon, A.B. (2015). Incidence of foot and mouth disease at polyclinics of SKUAST-Kashmir. Veterinary Science Research Journal. 6(2): 1-6.

    2. AlZubi, A.A. (2024). Detection of disease in calves using artificial intelligence. Indian Journal of Animal Research. 58(12): 2138-2145. doi: 10.18805/IJAR.BF-1760.

    3. Andrew, M., Fry, S.A. and Trotter, M.J. (2019). Automated detection and identification of individual cattle using deep learning and YOLOv2. Computers and Electronics in Agriculture. 165: 104959.

    4. Arora, R., Singh, V. and Sharma, R.K. (2019). Incidence and economic losses due to mastitis in dairy cattle in India: A review. Indian Journal of Animal Sciences. 89(9): 939- 948.

    5. Bhattacharya, S., Singh, R.K., Barman, N.N., Kumar, A., Chopra, M., Baitha, R., Upadhyay, C., Dubey, P.K. and Singh, K.P. (2023). Lumpy skin disease in India: Epidemiology and control strategies. Veterinary World. 16(1): 50-59.

    6. Chen, X., Li, Y. and Liu, H. (2024). Behaviour-YOLO: Cattle Behavior Recognition Based on Improved YOLO Network. In: Proceedings of SPIE. 13709: 1370916.

    7. DAHD (2019a). 20th Livestock Census of India-All India Report. Ministry of Fisheries, Animal Husbandry and Dairying, Government of India, New Delhi.

    8. DAHD (2019b). Key Results of 20th Livestock Census-2019. Government of India, New Delhi.

    9. DAHD (2019c). Basic Animal Husbandry Statistics. Government of India, New Delhi.

    10. Dominguez, E., Escarmis, C., Baranowski, E. et al. (2003). Evolution of foot-and-mouth disease virus. Virus Research. 91: 47-63.

    11. El-Henawy, M., Abdel-Aziz, H. and Abdel-Wahab, M. (2017). Cattle identification using convolutional neural networks and texture features. IET Computer Vision. 11(8): 667-675.

    12. FAO (2016). Economic Analysis of Animal Diseases. FAO Animal Production and Health Guidelines. Rome.

    13. FAO (2017). The Future of Livestock in India. FAO Animal Production and Health Paper. Rome.

    14. FAO (2018). World Livestock: Transforming the Livestock Sector Through the Sustainable Development Goals. Rome.

    15. FAO (2020). Buffalo Production and Health in South Asia. FAO Regional Office for Asia and the Pacific, Bangkok.

    16. FAO (2022). The State of Animal Genetic Resources in India. Rome.

    17. Govindaraj, G., Kumar, G.B., Krishnamohan, A. et al. (2020). Economic impact of foot-and-mouth disease in cattle and buffaloes in India. Indian Journal of Animal Sciences. 90(7): 971-976.

    18. Gupta, H., Jindal, P., Verma, O.P., Arya, R.K., Ateya, A.A., Soliman, N.F. and Mohan, V. (2022). Computer vision-based approach for automatic detection of dairy cow breed. Electronics. 11(22): 3791. https://doi.org/10.3390/ electronics11223791.

    19. Gupta, S.C., Tundup, T., Gupta, N. and Kumar, P. (1996). Livestock wealth of the Ladakh: A cold arid region in India. Animal Genetic Resources. 19: 27-36.

    20. Hamadani, H. (2013). Morphometry, Production Characteristics and Rearing Practices of Domestic Geese of Kashmir. M.V.Sc. Thesis. SKUAST-Kashmir, India.

    21. Hasan, M. and Mia, M.M. (2021). Prevalence and epidemiological risk factors of foot-and-mouth disease in cattle. Bulgarian Journal of Veterinary Medicine. 24: 1-14.

    22. He, K., Zhang, X., Ren, S. and Sun, J. (2016). Deep Residual Learning for Image Recognition. In: Proceedings of the IEEE CVPR. pp. 770-778.

    23. ICAR (2019). Indigenous Cattle and Their Conservation in India. Indian Council of Agricultural Research, New Delhi.

    24. ICAR-NBAGR (2020). Breed Descriptor of Ladakhi Cattle. National Bureau of Animal Genetic Resources, Karnal, India.

    25. ICAR-NIVEDI (2023). Animal Disease Informatics Report: Jammu and Kashmir. National Institute of Veterinary Epidemiology and Disease Informatics, Bengaluru, India.

    26. Iqbal, S. and Pampori, Z.A. (2008). Production potential and qualitative traits of indigenous chicken of Kashmir. Livestock Research for Rural Development. 20(11): 14.

    27. Kaggle Community (2023). Indian Bovine Breeds Dataset. Kaggle. https://www.kaggle.com/datasets/indian-bovine-breeds.

    28. Kappali, H., Sadyojatha, K.M. and Prashanthi, S.K. (2023). Computer vision and machine learning in paddy diseases identification and classification: A review. Indian Journal of Agricultural Research. 57(2): 183-187. doi: 10.18805/IJARe.A-6061.

    29. King, D.P., Ludi, A., Wilsden, G., Parida, S. and Paton, D.J. (2015). Differentiation of Vaccinated and Infected Animals Using Non-Structural Proteins. In: OIE Regional Commission Proceedings. pp. 1-4.

    30. Krishnamoorthy, P., Akshata, L.G., Jacob, S.S. et al. (2021b). Prevalence of theileriosis in cattle and buffaloes in India. Indian Journal of Animal Sciences. 91(4): 269-279.

    31. Krishnamoorthy, P., Ashwini, M., Suresh, K.P. and Roy, P. (2019a). Prevalence of Anaplasma species in dairy animals: A systematic review. Research in Veterinary Science. 123: 159-170.

    32. Krishnamoorthy, P., Goudar, A.L., Suresh, K.P. and Roy, P. (2021a). Global prevalence of mastitis in dairy cattle and buffaloes.  Research in Veterinary Science. 136: 561-586.

    33. Krishnamoorthy, P., Hamsapriya, S., Ashwini, M. et al. (2019b). Livestock-associated methicillin-resistant Staphylococcus aureus  in India. International Journal of Livestock Research. 9: 179-191.

    34. Krishnamoorthy, P., Suresh, K.P. and Roy, P. (2020). Meta-analysis as a tool for estimating prevalence of livestock diseases. Research and Reviews: Journal of Veterinary Sciences. 8: 1-10.

    35. Krishnamoorthy, P., Suresh, K.P., Saha, S. et al. (2017). Meta- analysis of prevalence of mastitis in dairy cattle in India. International Journal of Current Microbiology and Applied Sciences. 6: 1214-1234.

    36. Kumari, P., Bharti, V.K., Kumar, D., Mukesh, M. and Sharma, I. (2019). Haematological and biochemical profiling of Ladakhi cow. Indian Journal of Animal Sciences. 90(4): 527-531.

    37. Lawrence, W.R. (2002). The Valley of Kashmir. Gulshan Publishers, Srinagar, India.

    38. Li, Z., Chen, H., Zhang, Y. et al. (2022). A lightweight deep learning model for cattle face recognition. Computers and Electronics in Agriculture. 200: 107208.

    39. Mahato, S., Kumar, R., Choudhary, A. et al. (2025). Dairy DigiD: A keypoint-based deep learning system for individual dairy cow identification. Frontiers in Artificial Intelligence. 8: 1545247.

    40. Makhdoomi, D., Gazi, M., Nabi, S. and Ahmed, S. (2013). Morphometric studies on adult double-humped camels of Ladakh, India. Emirates Journal of Food and Agriculture. 25(7): 544- 548.

    41. Malik, R. and Sharma, R. (2019). Tick-borne diseases in buffaloes of northern India. Veterinary World. 12(6): 894-900.

    42. Mamatha, D.M., Venkata, L.P., Thanikonda, D.N., Kotagaram, H. and Koduru, S.K. (2026). Integrated DNA barcoding and QR code-based image recognition for identification of indian cattle breeds. Indian Journal of Animal Research. doi: 10.18805/IJAR.B-5745.

    43. Office of the Registrar General (2021). District-Wise Details of Livestock Population in Jammu and Kashmir-20th Livestock Census 2019. Open Government Data Platform India.

    44. Pawaskar, P., Yogish, H.K., Pakruddin, B. and Deepa, Y. (2025). Novel and efficient deep learning models for assessing AI’s impact on disease diagnosis in agriculture. Indian Journal of Agricultural Research. 59(9): 1395-1403. doi: 10.18805/IJARe.A-6388.

    45. Rahman, S.M., Singh, P.K. and Tiwari, A.K. (2023). Epidemiology and economic impact of lumpy skin disease in cattle in India. Tropical Animal Health and Production. 55: 219.

    46. Rather, A.F.K., Ganai, N.A., Shah, A.A. and Sheikh, A.A. (2022). Livestock and poultry breeds of Jammu and Kashmir and Ladakh. Indian Journal of Animal Sciences. 92(4): 409-416.

    47. Rather, A.F.K., Sheikh, A.A., Shah, A.A. et al. (2025). Computer vision in precision livestock farming. Computers and Electronics in Agriculture. 222: 110555.

    48. Rege, J.E.O. and Okeyo, M.A. (2006). Improving Our Knowledge of Tropical Indigenous Animal Genetic Resources. FAO/ AGTR Training Module Version 2. Rome.

    49. Saxena, A., Faiq, M., Ghatrehsamani, S. and Zahra, S.R. (2023). integrating artificial intelligence and machine learning for sustainable development in agriculture and allied sectors of the temperate himalayas. Agri. Engineering. 8(1): 35.

    50. Sharma, R.K. (2021). Fasciolosis in livestock of the Himalayan region. Journal of Parasitic Diseases. 45(3): 678-684.

    51. Sharma, S., Katoch, R. and Verma, N. (2020). Prevalence of major bovine diseases in Jammu and Kashmir: A retrospective study. Journal of Entomology and Zoology Studies. 8(3): 1230-1235.

    52. Shergojry, S.A., Akhoon, Z.A., Mubarak, T. and Namgyal, D. (2017). Socio-economic impact of livestock in tribal areas of Leh. Journal of Krishi Vigyan. 6(1): 187-190.

    53. Silva, J.E.P., Krohling, R.A. and Henriques, R.P.B. (2021). Computer vision-based systems for individual cattle identification: A review. Computers and Electronics in Agriculture. 185: 106131.

    54. Singh, A.K., Gupta, P.K. and Kumar, S. (2018). Prevalence and economic losses due to brucellosis in dairy animals in India. Indian Journal of Animal Sciences. 88(10): 1173- 1180.

    55. Tan, M. and Le, Q. (2021). EfficientNetV2: Smaller Models and Faster Training. In: Proceedings of the 38th ICML. 139: 10096-10106.

    56. Vijayalakshmi, A., Shanmugavadivu, P., Vijayalakshmi, S., Padarha, S. and Sivaranjani, R. (2024). Ensemble Learning Algorithm for Cattle Breed Identification using Computer Vision Techniques. In: Proceedings of the International Conference on Artificial Intelligence, Communication, IoT, Data Engineering and Security (IACIDS 2023). EAI. https:// doi.org/10.4108/eai.23-11-2023.2343338.

    57. Warhade, R., Devi, I., Singh, N., Arya, S., Dudi, K., Lathwal, S.S. and Tomar, D.S. (2024). Attention module incorporated transfer learning empowered deep learning-based models for classification of phenotypically similar tropical cattle breeds (Bos indicus). Tropical Animal Health and Production. 56(6): 192. https://doi.org/10.1007/s11250- 024-04050-7.

    58. WOAH (2021a). Foot and Mouth Disease (Infection with Foot and Mouth Disease Virus). WOAH Technical Disease Card. Paris.

    59. WOAH (2021b). Lumpy Skin Disease. WOAH Technical Disease Card. Paris.
    In this Article
      Contact us
      Follow us
      Published In
      Indian Journal of Animal Research

      Editorial Board

      View all (0)