Cattle image datasets for lameness detection and analysis

Cattle image datasets for detection and analysis of lameness are customized datasets collected in their rawness for training and validating machine learning and deep learning models, which are popularly used for detection and classification tasks. On September 9, 2025, we employed an EZVIZ WIFI mobile camera to collect 277 images of Wagyu and Angus cattle from Tau Sa farm in South Africa, capturing their body views. The cattle datasets were labeled appropriately. Folder 1 (containing label 1-277) represents the original datasets for both Wagyu and Angus cattle, while Folder 2 (containing label 276a-276e and label 277a-277e) represents the masked datasets for label 276 and label 277 datasets, which are considered as images that have excellent biological features. The datasets consist of active and inactive cattle, where standing and eating positions signify active, and lying down position signifies inactive, and probable weakness. The utilization of such datasets can speed up the training and validation of deep learning models for the development of automated livestock monitoring systems, whereby management efficiency and operational effectiveness are enhanced within the livestock industry. Moreover, integration of such biological information can assist specific models for cattle body condition scoring and health monitoring, enabling early identification and prevention of diseases, such as lameness. The datasets aim to boost public datasets for research purposes, fostering efficient and sustainable approaches for identifying and monitoring the health and productivity of cattle.