Fish recognition API
This is a sea fish classifier. It was made with 44,134 images corresponding to 44 species, which were scraped, or downloaded manually in some cases. It was done by transfer of learning using the RESNET 34 model from Pytorch and the Fastai2 libraries. In the validation of the model, the training error was 7%.