Feature Pyramid Networks for Object Detection

Note

A development version based on FPN.
Support multi-gpu training!

Abstract

This is a tensorflow re-implementation of Feature Pyramid Networks for Object Detection.

This project is based on Faster-RCNN, and completed by YangXue and YangJirui.

Train on VOC 2007 trainval and test on VOC 2007 test (PS. This project also support coco training.)

Comparison

use_voc2007_metric

Models	mAP	sheep	horse	bicycle	bottle	cow	sofa	bus	dog	cat	person	train	diningtable	aeroplane	car	pottedplant	tvmonitor	chair	bird	boat	motorbike
Faster-RCNN resnet50_v1	73.09	72.11	85.63	77.74	55.82	81.19	67.34	82.44	85.66	87.34	77.49	79.13	62.65	76.54	84.01	47.90	74.13	50.09	76.81	60.34	77.47
Faster-RCNN resnet101_v1	74.63	76.35	86.18	79.87	58.73	83.4	74.75	80.03	85.4	86.55	78.24	76.07	70.89	78.52	86.26	47.80	76.34	52.14	78.06	58.90	78.04
Faster-RCNN mobilenet_v2	50.34	46.99	68.45	65.89	28.16	53.21	46.96	57.80	38.60	44.12	66.20	60.49	52.40	56.06	72.68	26.91	49.99	30.18	39.38	38.54	64.74
FPN resnet50_v1	74.26	73.27	82.23	82.99	61.27	80.59	72.73	81.37	85.26	84.76	80.33	77.43	65.31	79.18	85.78	46.47	73.10	55.99	76.11	59.80	81.19
FPN resnet101_v1	76.14	74.63	85.13	81.67	63.79	82.43	77.83	83.07	86.45	85.82	81.08	81.01	71.22	80.01	86.30	48.05	73.89	56.99	78.33	62.91	82.24
FPN resnet101_v1+	75.71	74.83	83.55	82.47	65.49	77.85	71.74	80.98	86.61	87.14	81.02	77.76	71.26	79.82	86.78	51.64	77.45	56.12	79.44	60.55	81.69
FPN resnet101_v1++	75.89	76.05	84.22	80.29	63.21	83.04	78.69	81.81	86.61	85.61	79.75	79.78	71.27	80.33	86.24	49.03	76.81	56.32	78.51	60.37	79.91

+: SHARE_NET=False
++: SHORT_SIDE_LEN=800, FAST_RCNN_MINIBATCH_SIZE=512

COCO

Model	Backbone	Train Schedule	GPU	Image/GPU	FP16	Box AP(Mask AP)
Faster (ours)	R50v1-FPN	1X	1X TITAN Xp	1	no	36.1
Faster (ours)	R50v1-FPN	1X	4X TITAN Xp	1	no	36.1
Faster (Face++ & Detectron)	R50v1-FPN	1X	8X TITAN Xp	2	no	36.4
Faster (SimpleDet)	R50v1-FPN	1X	8X 1080Ti	2	no	36.5

My Development Environment

1、python3.5 (anaconda recommend)
2、cuda9.0 (If you want to use cuda8, please set CUDA9 = False in the cfgs.py file.)
3、opencv(cv2)
4、tfplot
5、tensorflow == 1.10

Download Model

Please download resnet50_v1、resnet101_v1 pre-trained models on Imagenet, put it to $PATH_ROOT/data/pretrained_weights.

Data Format

├── VOCdevkit
│   ├── VOCdevkit_train
│       ├── Annotation
│       ├── JPEGImages
│   ├── VOCdevkit_test
│       ├── Annotation
│       ├── JPEGImages

Compile

cd $PATH_ROOT/libs/box_utils/cython_utils
python setup.py build_ext --inplace

Demo(available)

Select a configuration file in the folder ($PATH_ROOT/libs/configs/) and copy its contents into cfgs.py, then download the corresponding weights.

cd $PATH_ROOT/tools
python inference.py --data_dir='/PATH/TO/IMAGES/' 
                    --save_dir='/PATH/TO/SAVE/RESULTS/' 
                    --GPU='0'

Eval

cd $PATH_ROOT/tools
python eval.py --eval_imgs='/PATH/TO/IMAGES/'  
               --annotation_dir='/PATH/TO/TEST/ANNOTATION/'
               --GPU='0'

Train

1、If you want to train your own data, please note:

(1) Modify parameters (such as CLASS_NUM, DATASET_NAME, VERSION, etc.) in $PATH_ROOT/libs/configs/cfgs.py
(2) Add category information in $PATH_ROOT/libs/label_name_dict/lable_dict.py     
(3) Add data_name to line 76 of $PATH_ROOT/data/io/read_tfrecord.py 

2、make tfrecord

cd $PATH_ROOT/data/io/  
python convert_data_to_tfrecord.py --VOC_dir='/PATH/TO/VOCdevkit/VOCdevkit_train/' 
                                   --xml_dir='Annotation'
                                   --image_dir='JPEGImages'
                                   --save_name='train' 
                                   --img_format='.jpg' 
                                   --dataset='pascal'

3、train

cd $PATH_ROOT/tools
python train.py

4、multi-gpu train

cd $PATH_ROOT/tools
python multi_gpu_train.py

Tensorboard

cd $PATH_ROOT/output/summary
tensorboard --logdir=.

Reference

1、https://github.com/endernewton/tf-faster-rcnn
2、https://github.com/zengarden/light_head_rcnn
3、https://github.com/tensorflow/models/tree/master/research/object_detection
4、https://github.com/CharlesShang/FastMaskRCNN
5、https://github.com/matterport/Mask_RCNN
6、https://github.com/msracver/Deformable-ConvNets