# FasterRCNN-pytorch
FasterRCNN is implemented in VGG, ResNet and FPN base.
reference:
rbg's FasterRCNN code: [https://github.com/rbgirshick/py-faster-rcnn](https://github.com/rbgirshick/py-faster-rcnn)
-----
# Model Performance
### Train on VOC2017 Test on VOC2017
| Backbone | mAp |
| ---------- |:------:|
| VGG16 | 0.7061 |
| ResNet101 | 0.754 |
# Train Your Model
### 1.Before Run You Need:
1. cd ./lib
> Change gpu_id in make.sh and setup.py.
Detially, you need modify parameter setting in line 5, 12 and 19 in make.sh and line 143 in setup.py where include key words '-arch=' depend on your gpu model.(select appropriate architecture described in table below)
> sh make.sh
| GPU model | Architecture |
| -------- | :-----: |
| TitanX (Maxwell/Pascal) | sm_52 |
| GTX 960M | sm_50 |
| GTX 108 (Ti) |sm_61 |
| Grid K520 (AWS g2.2xlarge) |sm_30 |
| Tesla K80 (AWS p2.xlarge) |sm_37 |
2. cd ../
> mkdir ./data
> mkdir ./data/pretrained_model
> download pre-trained weights in ./data/pretrained_model
3. run train.py
### 2.How to use?
#### **Note: decentralization in preprocesing is based on BGR channels, so you must guarantee your pre-trained model is trained on the same channel set if you use transfer learning**
For example:
VGG:
CUDA_VISIBLE_DEVICES=1 python train.py --net='vgg16' --tag=vgg16 --iters=70000 --cfg='./experiments/cfgs/vgg16.yml' --weight='./data/pretrained_model/vgg16_caffe.pth'
CUDA_VISIBLE_DEVICES=2 python test.py --net='vgg16' --tag=vgg16 --model=60000 --cfg='./experiments/cfgs/vgg16.yml' --model_path='voc_2007_trainval/vgg16/vgg16_faster_rcnn' --imdb='voc_2007_test' --comp
ResNet:
CUDA_VISIBLE_DEVICES=2 python train.py --net='res18' --tag=res18 --iters=70000 --cfg='./experiments/cfgs/res18.yml' --weight='./data/pretrained_model/Resnet18_imagenet.pth'
CUDA_VISIBLE_DEVICES=3 python train.py --net='res50' --tag=res50 --iters=70000 --cfg='./experiments/cfgs/res50.yml' --weight='./data/pretrained_model/Resnet50_imagenet.pth'
CUDA_VISIBLE_DEVICES=7 python train.py --net='res101' --tag=res101 --iters=80000 --cfg='./experiments/cfgs/res101.yml' --weight='./data/pretrained_model/resnet101_caffe.pth'
CUDA_VISIBLE_DEVICES=6 python test.py --net='res101' --tag=res101_1 --cfg='./experiments/cfgs/res101.yml' --model=70000 --model_path='voc_2007_trainval/res101_1' --imdb='voc_2007_test' --comp
----
FasterRCNN-pytorch-master.zip （86个子文件） 
FasterRCNN-pytorch-master experiments cfgs 
res101.yml 344B vgg16.yml 357B res101-lg.yml 449B fpn101.yml 543B mobile.yml 324B fpn50.yml 543B res18.yml 337B res50.yml 322B logs .gitignore 8B log 1KB LICENSE 1KB README.md 2KB test.py 5KB lib setup.py 6KB nets vgg16.py 3KB __init__.py 0B layers_util.py 3KB network.py 16KB fpn.py 11KB network_fpn.py 22KB resnet.py 8KB model nms_wrapper.py 930B bbox_transform.py 2KB train_val.py 17KB __init__.py 0B config.py 11KB test.py 8KB train_val_ori.py 19KB bbox_transform_cpu.py 2KB datasets __init__.py 0B pascal_voc.py 11KB imdb.py 9KB voc_eval.py 7KB factory.py 2KB rpn proposal_target_layer.py 8KB __init__.py 0B anchor_target_layer_gpu.py 7KB generate_anchors_global.py 1KB fpn proposal_target_layer.py 8KB __init__.py 0B generate_anchors_global.py 2KB anchor_target_layer_cpu.py 8KB proposal_layer.py 3KB anchor_target_layer_cpu.py 6KB generate_anchors.py 3KB proposal_layer.py 3KB nms src cuda nms_kernel.cu 3KB nms_kernel.h 341B nms_cuda.h 108B nms.h 155B nms_cuda.c 2KB nms.c 3KB __init__.py 0B pth_nms.py 1KB nms_test .gitignore 15B __init__.py 0B nms_kernel.cu 5KB gpu_nms.pyx 1KB gpu_nms.hpp 146B py_cpu_nms.py 1KB cpu_nms.pyx 2KB build.py 774B __init__.py 0B roi_pooling roi_pool.py 2KB src roi_pooling.h 178B cuda roi_pooling_kernel.cu 8KB roi_pooling_kernel.h 767B roi_pooling.c 4KB roi_pooling_cuda.h 420B roi_pooling_cuda.c 3KB __init__.py 0B build.py 822B roi_pool_py.py 2KB _ext __init__.py 0B roi_pooling __init__.py 310B roi_data_layer __init__.py 248B roidb.py 2KB layer.py 3KB minibatch.py 3KB utils __init__.py 0B blob.py 1KB timer.py 948B bbox.py 1KB bbox.pyx 3KB make.sh 460B train.py 6KB
评论0