• Architecure

  • (alexnet) ImageNet Classification with Deep Convolutional Neural Networks
  • (googlenet) Going Deeper with Convolutions
  • (vgg) Very Deep Convolutional Networks for Large-Scale Image Recognition
  • (resnet) Deep Residual Learning for Image Recognition
  • (sppnet) Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition
  • (mobilenet_v1) MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
  • (yolo_v1) You Only Look Once: Unified, Real-Time Object Detection
  • (yolo_v2) YOLO9000: Better, Faster, Stronger

• Python >= 3.5
• PyTorch >= 1.0.1.post2
• torchvision >= 0.2.2.post3
• TensorFlow/Tensorboard (if you want to use the tensorboard for visualization)
• Other dependencies (pyyaml, easydict, tensorboardX)

pip install -r requirements.txt

simply run the cmd for the training:

## 1 GPU for vgg
CUDA_VISIBLE_DEVICES=0 python -u train_classifier.py --work-path ./experiments/vgg/cifar10

## resume from ckpt
CUDA_VISIBLE_DEVICES=0 python -u train_classifier.py --work-path ./experiments/vgg/cifar10 --resume

## 2 GPUs for inception_v1
CUDA_VISIBLE_DEVICES=0,1 python -u train_classifier.py --work-path ./experiments/inception_v1/cifar10

## 4 GPUs for yolo_v2
CUDA_VISIBLE_DEVICES=0,1,2,3 python -u train_detector.py --work-path ./experience/yolo_v2/voc2007

We use yaml file config.yaml to save the parameters, check any files in ./experimets for more details.
You can see the training curve via tensorboard, tensorboard --logdir path-to-event --port your-port. (Not verified)
The training log will be dumped via logging, check log.txt in your work path.