athanzxyt/resnet

Re-implementation of ResNet in PyTorch

ResNet in PyTorch

This is a re-implementation of the ResNet model into PyTorch. Please note that ResNet architecture for the ImageNet dataset found in the paper was implemented. However, was tested on the CIFAR-10 dataset.

Training on NVIDIA RTX A6000 yielded the following results:

	ResNet18	ResNet50
Top1 Accuracy	87.5%	81.3%
Top5 Accuracy	98.4%	99.2%

As mentioned before, ResNet18 and ResNet50 were originally made for the ImageNet dataset. The authors of the papers did create other ResNet infrastructures for CIFAR, namely Resnet32 and others.

Here is a summary of the files in this respository:

File Name	Description
resnet.py	ResNet class
main.py	Script to run tests on the CIFAR dataset
accXX.png	Accuracy graph of Classification over Iteration.
image.png	Sample of CIFAR dataset

Citations

[1] He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). https://doi.org/10.1109/cvpr.2016.90