Deep learning based 2D computer vision in precision livestock farming for pigs: A systematic literature review and future research directions

The pressing challenges in pig production - particularly in welfare monitoring, behavior analysis, and health management - have driven increasing research efforts in computer vision as a scalable and automated solution. Over the past decade, a diverse body of work has emerged within precision livestock farming (PLF), addressing tasks ranging from pig detection and identification to behavior recognition and health assessment. This expanding research landscape calls for a structured synthesis to consolidate existing knowledge, clarify methodological developments, and elucidate the role of behavior-focused applications. This systematic literature review examines the adoption of computer vision in pig production through four research dimensions: (i) the range of problems addressed, (ii) the models and neural network architectures employed, (iii) the evaluation metrics reported, and (iv) the technical and practical challenges associated with developing and deploying such models. The contribution of this review is threefold. First, it offers an accessible conceptual overview of key deep learning based computer vision approaches to support interdisciplinary researchers entering the field. Second, it synthesizes methodological and evaluative trends across studies. Third, it identifies key challenges and research opportunities for advancing robust and practically relevant computer vision systems in pig farming.