A simple convolutional neural network made using pytorch for Human-Computer-Communication course, the net scored 80% accuracy after training on the Intel dataset. My object was to implement image classification in any method I desire.

Intel Image Classification

Just run the image_recognizer.py

If you desire to train your own model please uncomment the following line (training may take around 10-20 minutes depending on your machine and if you enabled CUDA acceleration)

#train_new_model(classes, dataloaders, 31, name='saved_model.pth')

If you desire to use pretrained model please use the following line and comment the line above

load_run(dataloaders[2], classes,'saved_model.pth')

I dont have a CUDA GPU so I cant really use it, if you desire to speed up your training please refer to pytroch documentation about CUDA