We built an image classifier to identify different genuses of dinosaurs. We wanted to do this with a CNN, as most of us didn't have any experience in it and we wanted to learn. It was a significant challenge for us and we ended up running into multiple obstacles, including the limited dataset and the GPU issues. None of us had an NVIDIA GPU so we ran into issues with the test dataset-we overcame this but using datahub to run our code instead. We also built a figma app prototype to accompany the classifier. We also wrote a significant amount of code on a GAN for image generation but because of its long training time we had to abandon this approach.
Log in or sign up for Devpost to join the conversation.