No Data
Paper: Aggregated Residual Transformations for Deep Neural Networks
This code takes ImageNet dataset as example. You can download ImageNet dataset and put them as follows. I only provide
ILSVRC2012_dev_kit_t12due to the restriction of memory, in other words, you need download
ILSVRC2012_img_trainand
ILSVRC2012_img_val.
├── train.py # train script ├── resnext.py # network of resnext ├── read_ImageNetData.py # ImageNet dataset read script ├── ImageData # train and validation data ├── ILSVRC2012_img_train ├── n01440764 ├── ... ├── n15075141 ├── ILSVRC2012_img_val ├── ILSVRC2012_dev_kit_t12 ├── data ├── ILSVRC2012_validation_ground_truth.txt ├── meta.mat # the map between train file name and label
python train.py --batch-size 256 --gpus 0,1,2,3
epoch_4.pth.tar, the
--start-epochparameter is corresponding to the epoch of the checkpoint):
python train.py --batch-size 256 --gpus 0,1,2,3 --resume output/epoch_4.pth.tar --start-epoch 4