Abstract
Numerous recent papers based on deep learning (DL) have been published covering a wide range of applications to pig production. These applications provide information susceptible of being used to make better decisions. However, the potential use as tools for supporting pig production decisions or the integration in existing or new decision models have not been explored yet.
The goal of this systematic literature review (SLR) is to provide an overview of recent developments in cutting-edge DL methodologies proposed in pig production and how they can serve to improve decision making processes.
The revised papers are analyzed under different dimensions: (1) authors and research institutions that have made the biggest contributions to DL for image processing, computer vision and other innovative applications in pig farms; (2) coverage of the echelons in the pig supply chain (3) technical aspects like data collection techniques, DL models, DL backbones, graphics processing units (GPUs), and evaluation metrics and (4) value of information.
The review is briefly extended to DL applications in other livestock species not yet present in pig production to enrich the discussion. The revised applications suggest that DL is mostly applied to automatize data gathering and processing and to monitor animals or on farm activities.
The current challenges and future research agenda are also identified envisioning the integration of DL and operational research(OR) methods as a way to produce more efficient decision-making support tools for the pig industry.