This is a repositoy to gather models implemented in caffe. Including classification, detection, semantic segmentation and human pose estimation
- Classification
<1> VGG-16, VGG-19
VGG-16: https://gist.github.com/ksimonyan/211839e770f7b538e2d8
VGG-19: https://gist.github.com/ksimonyan/3785162f95cd2d5fee77#file-readme-md
<2> ResNet( ResNet-50, ResNet-101, ResNET-151)
Official project: https://github.com/KaimingHe/deep-residual-networks
<3> MobileNet
Third party project: https://github.com/shicai/MobileNet-Caffe
Official project (tensorfow) :https://github.com/tensorflow/models/blob/master/slim/nets/mobilenet_v1.md
<4> DenseNet
Official project: https://github.com/liuzhuang13/DenseNetCaffe
<5> SqueezeNet
official project: https://github.com/DeepScale/SqueezeNet
- Detection
<1> Faster R-CNN
Official project:
<2> R-FCN
Official project:
<3> SSD
Official project: https://github.com/weiliu89/caffe/tree/ssd
- Semantic Segmenation
<1> FCN
Official project: https://github.com/shelhamer/fcn.berkeleyvision.org
Main contribution: fully convolutional network for dense prediction; Encode-Decode structure
<2> SegNet
Official project: https://github.com/alexgkendall/SegNet-Tutorial.git
Main contribution: symmetric encode-decode structure
<3> Deeplab-v1
Official project: https://bitbucket.org/deeplab/deeplab-public/
Main contribution: dilate convoliton + CRF
<4> Deeplab-v2
Official project: https://bitbucket.org/aquariusjay/deeplab-public-ver2
Main contribution: ResNet-101 + ASPP
<5> Deeplab-v3
- Human Pose estimation
<1> Hourglass
<2> CMU-Pose (OpenPose)
Official project :https://github.com/ZheC/Realtime_Multi-Person_Pose_Estimation
Main contribution: Bottom-up structure, multi-task learning (keypoints + PAF)
<3> CPM (Convolutional Pose Machine)
Official project: https://github.com/shihenw/convolutional-pose-machines-release
Main contribution: multi-stage refinement