PVANet
by Sanghoon Hong, Byungseok Roh, Kye-hyeon Kim, Yeongjae Cheon, Minje Park (Intel Imaging and Camera Technology) Presented in EMDNN2016, a NIPS2016 workshop (arXiv link)
This repository is a fork from py-faster-rcnn and demonstrates the performance of PVANet.
You can refer to py-faster-rcnn README.md and faster-rcnn README.md for more information.
Please note that this repository doesn't contain our in-house codes used in the published article. - This version of py-faster-rcnn is slower than our in-house runtime code (e.g. image pre-processing code written in Python) - PVANet was trained by our in-house deep learning library, not by this implementation. - There might be a tiny difference in VOC2012 test results, because some hidden parameters in py-faster-rcnn may be set differently with ours.
If you want to cite this work in your publication:
@article{hong2016pvanet, title={{PVANet}: Lightweight Deep Neural Networks for Real-time Object Detection}, author={Hong, Sanghoon and Roh, Byungseok and Kim, Kye-Hyeon and Cheon, Yeongjae and Park, Minje}, journal={arXiv preprint arXiv:1611.08588}, year={2016} }
Install Dependency
sudo pip install Cython
Clone the Faster R-CNN repository
# Make sure to clone with --recursivegit clone --recursive https://github.com/sanghoon/pva-faster-rcnn.git
We'll call the directory that you cloned Faster R-CNN into FRCN_ROOT
. Build the Cython modules
cd $FRCN_ROOT/lib make
Build Caffe and pycaffe
cd $FRCN_ROOT/caffe-fast-rcnn# Now follow the Caffe installation instructions here:# http://caffe.berkeleyvision.org/installation.html# For your Makefile.config:# Uncomment `WITH_PYTHON_LAYER := 1`cp Makefile.config.example Makefile.config make -j8 && make pycaffe
Download PVANet detection model for VOC2007
cd $FRCN_ROOT./models/pvanet/download_voc2007.sh
Download PVANet detection model for VOC2012 (published model)
cd $FRCN_ROOT./models/pvanet/download_voc_best.sh
(Optional) Download all available models (including pre-trained and compressed models)
cd $FRCN_ROOT./models/pvanet/download_all_models.sh
(Optional) Download ILSVRC2012 (ImageNet) classification model
cd $FRCN_ROOT./models/pvanet/download_imagenet_model.sh
(Optional) If the scripts don't work, please download the models from ...
| Model | Google Drive | | ------ | ---- | | PVANet for VOC2007 | link | | PVANet for VOC2012 | link | | PVANet for VOC2012 (compressed) | link | | PVANet for ILSVRC2012 (ImageNet) | link | | PVANet pre-trained | link |
Download PVANet detection models
Run the demo script
cd $FRCN_ROOT./tools/test_pvanet.py
Download PASCAL VOC 2007 and 2012 -- Follow the instructions in py-faster-rcnn README.md
PVANet on PASCAL VOC 2007
cd $FRCN_ROOT./tools/test_net.py --net models/pvanet/pva9.1/PVA9.1_ImgNet_COCO_VOC0712.caffemodel --def models/pvanet/pva9.1/faster_rcnn_train_test_21cls.pt --cfg models/pvanet/cfgs/submit_1019.yml --gpu 0
PVANet (compressed)
cd $FRCN_ROOT./tools/test_net.py --net models/pvanet/pva9.1/PVA9.1_ImgNet_COCO_VOC0712plus_compressed.caffemodel --def models/pvanet/pva9.1/faster_rcnn_train_test_ft_rcnn_only_plus_comp.pt --cfg models/pvanet/cfgs/submit_1019.yml --gpu 0
| Model | VOC2007 mAP (%) | VOC2012 mAP (%) | | --------- | ------- | ------- | | PVANet+ (VOC2007) | 84.9 | N/A | | PVANet+ (VOC2012) | 89.8 | 84.2 | | PVANet+ (VOC2012 + compressed) | 87.8 | 83.7 | - The training set for the VOC2012 model includes the VOC2007 test set. Therefore the accuracies on VOC2007 of the model are not meaningful; They're shown here just for reference
| Input size | Top-1 error (%) | Top-5 error (%) | | --- | --- | --- | | 192x192 | 30.00 | N/A | | 224x224 | 27.66 | 8.84 | - We re-trained a 224x224 model from the '192x192' model as a base model.
上一篇:probtorch
还没有评论,说两句吧!
热门资源
Keras-ResNeXt
Keras ResNeXt Implementation of ResNeXt models...
seetafaceJNI
项目介绍 基于中科院seetaface2进行封装的JAVA...
spark-corenlp
This package wraps Stanford CoreNLP annotators ...
capsnet-with-caps...
CapsNet with capsule-wise convolution Project ...
inferno-boilerplate
This is a very basic boilerplate example for pe...
智能在线
400-630-6780
聆听.建议反馈
E-mail: support@tusaishared.com