Increasing piglet survival remains a priority for pig producers. In breeding programs, it can be increased by considering piglet maturity at birth, defined as the level of full development. Indeed, an immature piglet, which does not reach its intrauterine growth, has a lower chance of surviving during the first few days after birth. On the farm, it is possible to detect these piglets because they present an asymmetrical growth with a specific head morphology. However, it is necessary to record this phenotype routinely to use this new criterion in breeding programs. The aim of this study was to develop a deep learning algorithm to predict piglet maturity automatically using pictures of their head. A first step was to record maturity levels of piglets at birth using, on the one hand, morphological criteria such as head shape, bulging eyes, ear shape and, on the other hand, morphological measurements i.e. body length, head length and skull length. Each piglet head was then photographed from the side and from the front. From 1,100 pictures of piglet heads annotated with their maturity level, a prediction model was developed and validated using an additional 661 pictures. Currently, the algorithm can detect an immature piglet with a sensitivity of 89% and a specificity of 74%. These are encouraging initial results, and improvement is expected by increasing the size of the database. This new system has great potential to record piglet maturity automatically and provide new perspectives for genetic selection.