https://github.com/Angzz/fpn-gluon-cv

Feature Pyramid Network for Object Detection
https://github.com/Angzz/fpn-gluon-cv

gluon-cv mxnet object-detection

Last synced: 2 months ago
JSON representation

Feature Pyramid Network for Object Detection

• Host: GitHub
• URL: https://github.com/Angzz/fpn-gluon-cv
• Owner: Angzz
• License: apache-2.0
• Created: 2019-01-16T10:08:32.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2022-11-16T10:05:38.000Z (about 2 years ago)
• Last Synced: 2024-08-01T22:40:09.393Z (5 months ago)
• Topics: gluon-cv, mxnet, object-detection
• Language: Python
• Homepage:
• Size: 3.04 MB
• Stars: 9
• Watchers: 3
• Forks: 3
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• Awesome-MXNet - FPN-gluon-cv

README

        
# Feature Pyramid Network for Object Detection

Paper Link : https://arxiv.org/pdf/1612.03144.pdf

### Note : Our code has been merged into official [gluon-cv](https://github.com/dmlc/gluon-cv), we recommend the official version, details can be found: https://github.com/dmlc/gluon-cv/pull/494

This repo is an unofficial implementation of FPN in a [gluon-cv](https://github.com/dmlc/gluon-cv) style, we will follow up with the official code and please refer to the official tutorials for more information.

# Usage

1.Install the mxnet 

* pip install mxnet-cu92 --pre --upgrade

2.Clone this code and run the script

* 

# Results

### VOC07

* **e2e_fastercnn_fpn_resnet50_v1b_1x_voc0712** : mAP=0.8035 on VOC07 Test.

  * train-log : https://drive.google.com/open?id=1obMRJedBHXrNPZLsahiR6WMQyYJFnli0

  * models : https://pan.baidu.com/s/1JrQYMLCPXuWGViSokS1LLg

* **e2e_fastercnn_fpn_resnet101_v1d_1x_voc0712** : mAP=0.8301 on VOC07 Test.

  * train-log : https://drive.google.com/open?id=1Z7FeZ7fEuhYi1d4ExBlaLcJLl8RBSfFG

### VOC12

* **e2e_fastercnn_fpn_resnet50_v1d_1x_voc07++12** : Stay tuned!

  * train-log : https://drive.google.com/open?id=18NFrME-TM_8cvYTkKwtrqvcPCmZ5L-MG

### COCO17

* **e2e_fastercnn_fpn_resnet50_v1d_2x_coco17** : Stay tuned!

# Acknowledgements

Below repos give me a lot of inspiration

* https://github.com/msracver/Deformable-ConvNets

* https://github.com/guoruoqian/FPN_Pytorch

* https://github.com/facebookresearch/Detectron