This is the source code of our TCSVT 2019 paper "Fast Fine-grained Image Classification via Weakly Supervised Discriminative Localization" , Please cite the following paper if you use our code.
Xiangteng He, Yuxin Peng and Junjie Zhao, "Fast Fine-grained Image Classification via Weakly Supervised Discriminative Localization", IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), Vol. 29, No. 5, pp. 1394-1407, May. 2019.
Our code is based on early version of Faster R-CNN in MXNet, all the dependencies are the same as it.
Here we use Cars-196 dataset for an example, we have organized the data as the form of PASCAL VOC dataset, which can be downloaded from link.
Download VGG16 pretrained model vgg16-0000.params from MXNet model gallery to model folder.
Start training and tesing by executiving the following commands. This will train and test the network on Cars-196 train.
- sh train_end2end_cam_prepare.sh
- python train_end2end_cam.py --gpu GPUID
- python test.py --gpu GPUID
- sh train_end2end_conv5_prepare.sh
- python train_end2end_conv5.py --gpu GPUID
- python test.py --gpu GPUID
- sh train_end2end_conv4_prepare.sh
- python train_end2end_conv4.py --gpu GPUID
- python test.py --gpu GPUID
Download the models that we trained from the link and unzipped to ./model/ foldes. For more information, please refer to our TCSVT paper.
Welcome to our Laboratory Homepage for more information about our papers, source codes, and datasets.