GesilaA/YOLO-FSM

Project of “Feature Selection Module for CNN Based Object Detector”

note

This project support YoloV3, YoloV4 and YoloV5.

install

pip install -r requirements.txt -i https://pypi.douban.com/simple

Environment

• Nvida GeForce TitanX
• CUDA10.0
• CUDNN7.0
• ubuntu 18.04
• python 3.6

pretrained weights

Baidu Netdisk: https://pan.baidu.com/s/1kxuHjR7qZTwGN3Z1y02AFw 9xja

data prepare

|${folder}
 |----YOLO-FSM
 |----weights
 |----dataset
      |----VOCtest-2007
           |----VOCdevkit
           |    |----VOC2007
           |         |----Annotations
           |         |----JPEGImages
           |         |----ImageSets
           |    
      |----VOCtrainval-2007
           |----VOCdevkit
           |    |----VOC2007
           |         |----Annotations
           |         |----JPEGImages
           |         |----ImageSets
           | 
      |----VOCtrainval-2012
           |----VOCdevkit
           |    |----VOC2012
           |         |----Annotations
           |         |----JPEGImages
           |         |----ImageSets
           | 

Train or Test

please set yolo model in train.py or test.py

cd utils; python voc.py; cd -
python train.py --weight_path ${weight_path}
python test.py --weight_path ${weight_path}

Result on Pascal VOC2007-test

Name	mAP	aero	bike	bird	boat	bottle	bus	car	cat	chair	cow	table	dog	horse	motor	person	plant	sheep	sofa	train	monitor
FasterRCNN	73.2	76.5	79.0	70.9	65.5	52.1	83.1	84.7	86.4	52.0	81.9	65.7	84.8	84.6	77.5	76.7	38.8	73.6	73.9	83.0	72.6
SSD300	74.3	75.5	80.2	72.3	66.3	47.6	83.0	84.2	86.1	54.7	78.3	73.9	84.5	85.3	82.6	76.2	48.6	73.9	76.0	83.4	74.0
HyperNet	76.3	77.4	83.3	75.0	69.1	62.4	83.1	87.4	87.4	57.1	79.8	71.4	85.1	85.1	80.0	79.1	51.2	79.1	75.7	80.9	76.5
CenterNet	81.5	89.6	89.0	79.3	73.3	75.3	86.6	89.6	86.6	67.0	87.1	75.1	85.6	90.0	87.0	86.2	58.7	80.5	73.9	87.9	81.6
YoloV3	81.0	88.0	87.4	81.8	69.6	73.7	85.7	88.6	87.0	67.2	85.7	74.6	87.1	87.7	84.9	85.6	58.9	84.1	76.1	84.6	81.1
YoloV3-SE	81.2	89.5	86.7	80.8	73.7	69.9	86.8	89.0	88.1	66.5	85.8	74.2	86.6	89.2	86.2	85.1	57.9	86.5	75.1	87.5	79.1
YoloV3-FSM	81.6	87.3	88.2	82.2	73.6	69.9	85.9	88.9	87.2	68.6	87.6	76.7	86.8	89.8	86.0	85.7	57.3	86.4	77.9	85.8	81.1
YoloV4	83.9	90.1	89.1	83.3	76.7	76.4	90.2	89.5	88.6	73.0	89.4	80.4	87.6	90.1	88.4	87.9	57.7	85.9	82.2	87.6	83.8
YoloV4-FSM	85.0	90.1	89.9	84.7	78.9	78.3	90.0	89.9	88.9	74.7	89.3	82.8	87.8	90.4	89.5	88.4	62.4	87.2	82.0	88.5	87.0
YoloV5l	80.7	89.6	87.2	78.3	71.5	74.9	86.1	89.4	86.1	64.8	86.6	75.5	84.3	90.0	85.6	85.6	56.3	83.5	75.5	85.2	78.2
YoloV5l-FSM	82.2	86.6	88.6	82.4	76.5	75.6	87.0	89.4	87.9	66.1	87.2	78.0	85.9	88.1	88.3	86.6	58.6	87.4	74.7	87.7	80.4

Visualization on VOC2007 Test set

Baidu Netdisk: https://pan.baidu.com/s/1FySdBBbZzkjyKr0KN9JpSQ hn8t