Image classification techniques

Evaluation of transfer learning and data augmentation in AlexNet and ResNet for image classification on Caltech-101. Experiments are carried out in PyTorch. More information is available in the PDF report.

Structure

• notebook.ipynb contains the main logic of the program and the results of the experiments that we carried out.

• The data folder contains the Caltech-101 dataset along with train and test indices, train.txt and test.txt, as well as the dataset handler class dataset.py, with methods to initialize and handle the dataset and to perform a stratified split for training and validation sets.

• manager.py contains the network manager class, which handles training, validation and testing of a neural network, as well as logging results such as training and validation loss and accuracy over epochs.

References

[1] Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. ImageNet classification with deep convolutional neural networks. Communications of the ACM, 2017.

[2] Li Fei-Fei, Rob Fergus, and Pietro Perona. Learning Generative Visual Models from Few Training Examples: An Incremental Bayesian Approach Tested on 101 Object Categories. IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, 2004.

[3] Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. ImageNet: A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009.

[4] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep Residual Learning for Image Recognition. 2015.