August 2024
Intermediate to advanced
374 pages
8h 44m
English
While a model that predicts cats and dogs with around 80% accuracy is pretty good, there’s still a lot of room to improve. The winning model for this exact dataset achieved 98% accuracy in 2013. Since then, there have been significant advances in the field of computer vision, and you have yet to exhaust all of the training tricks in the book. In this section, you’ll make a few small tweaks to your model and training process, and you’ll see how they help you push your model toward 90% accuracy.
Right now you only have about 25,000 images total for training. For a typical deep learning problem, that’s not a lot. You could go out and collect more labeled images of cats and dogs, but that’s tedious ...
Read now
Unlock full access