johndan070 / resnet_finetune_cub

Fine tuning Codes for ResNet on cub-200-2011

Fine Tune ResNet on CUB-200-2011 and Stanford Cars Datasets

Introduction

This repo contains codes for fine tuning ResNet on CUB_200_2011 datasets.

Because ResNet_SE and ResNet_ED's model files do not belong to me, so I remove them in the projects.

The ResNet models provided by torchvision are available.

Datasets

1.CUB200-2011

  CUB-200-2011 dataset has 11,788 images of 200 bird species. The project page is as follows.

  Detailed information as follows:

• Images are contained in the directory data/cub200/raw/images/, with 200 subdirectories.
• Format of images.txt: <image_id> <image_name>
• Format of train_test_split.txt: <image_id> <is_training_image>
• Format of classes.txt: <class_id> <class_name>
• Format of iamge_class_labels.txt: <image_id> <class_id>

2.Stanford Cars

Stanford Cars datasets has 16185 images of 196 car species. The project page is as follows.

Stanford Car

Detailed information as follows:

• Directory car_ims contains total images (both training and testing images, whose number is 16185)
• File car_nori.list contains information as follows:

How to use

git clone https://github.com/zhangyongshun/resnet_finetune_cub.git
cd base_model_finetune
python train.py [params]

Results

There are some results as follows: