Classification on cifar10 dataset

This is a simple project of image classification. I trained my CNN model with cifar10 dataset using GPU T4 x2 from kaggle. You can check my kaggel notebook here. My model has top1 accuracy of 71% and top5 accuracy of 97%. path to the trained model is :src/models/

Try this classification model

As my model is not perfect it'll make faulty predictions sometimes. If you want better classification model you can check this repo.

Requirements

• numpy, tensorflow

You can try my Classification model(pretrained models) by using following code

# in repo's root directory
python3 run.py --idx image_idx

• image_idx : cifar10 contains 10000 test samples. So you can choose any integer between 0 and 9999.

example

python3 run.py --idx 45

output

Accuray and Loss