Accurate identification of individual cows is essential to precision dairy farming. While computer vision offers a non-invasive alternative to ear tags and RFID systems, its practical deployment remains limited by the need for zero-shot identification in dynamic herds where test identities are unseen during training. In this work, we propose OpenCowID, a unified framework that addresses this challenge. First, we introduce a stochastic cow coat synthesis pipeline that efficiently generates large-scale, diverse images. Second, using the generated large-scale high-quality data, we present a centroid-guided feature learning strategy that forms a well-structured embedding space using virtual class centroids, enabling generalization to unseen identities. OpenCowID achieves state-of-the-art zero-shot and open-set identification on real-world cow benchmarks, without requiring any real labeled training data. This work contributes to the advancement of automated livestock monitoring, enabling robust, non-invasive identification. Code is available at https://github.com/neis-lab/OpenCowID.

Precision livestock farming (PLF) has been transformed by machine learning (ML), enabling more precise and timely interventions that enhance overall farm productivity, animal welfare, and environmental sustainability. However, despite the availability of various sensing technologies, few datasets leverage multiple modalities, which are crucial for developing more accurate and efficient monitoring devices and ML models. To address this gap, we present MmCows, a multimodal dataset for dairy cattle monitoring. This dataset comprises a large amount of synchronized, high-quality measurement data on behavioral, physiological, and environmental factors. It includes two weeks of data collected using wearable and implantable sensors deployed on ten milking Holstein cows, such as ultra-wideband (UWB) sensors, inertial sensors, and body temperature sensors. In addition, it features 4.8 million frames of high-resolution image sequences from four isometric view cameras, as well as temperature and humidity data from environmental sensors. We also gathered milk yield data and outdoor weather conditions. One full day’s worth of image data is annotated as ground truth, totaling 20,000 frames with 213,000 bounding boxes of 16 cows, along with their 3D locations and behavior labels. An extensive analysis of MmCows is provided to evaluate the modalities individually and their complementary benefits. The release of MmCows and its benchmarks will facilitate research on multimodal monitoring of dairy cattle, thereby promoting sustainable dairy farming. The dataset and the code for benchmarks are available at https://github.com/neis-lab/mmcows.