keras-yolo3-Mobilenet

Introduction

A Keras implementation of YOLOv3 (Tensorflow backend) inspired by allanzelener/YAD2K.

And I change the backend of darknet53 into

• Mobilenet
• VGG16
• ResNet101
• ResNeXt101

Experiment on open datasets

Model name	InputSize	TrainSet	TestSet	mAP	Speed	Ps
YOLOv3-Mobilenet	320x320	VOC07	VOC07	64.22%	29fps	Keras on 1080Ti
YOLOv3-Mobilenet	320x320	VOC07+12	VOC07	74.56%	29fps	Keras on 1080Ti
YOLOv3-Mobilenet	416x416	VOC07+12	VOC07	76.82%	25fps	Keras on 1080Ti
MobileNet-SSD	300x300	VOC07+12+coco	VOC07	72.7%	(unknown)
MobileNet-SSD	300x300	VOC07+12	VOC07	68%	(unknown)
Faster RCNN, VGG-16	~1000x600	VOC07+12	VOC07	73.2%	151ms	Caffe on Titan X
SSD,VGG-16	300x300	VOC07+12	VOC07	77.5%	39fps	Keras on Titan X

PS:

1. Compared with MobileNet-SSD, YOLOv3-Mobilenet is much better on VOC2007 test, even without pre-training on Ms-COCO
2. I use the default anchor size that the author cluster on COCO with inputsize of 416*416, whereas the anchors for VOC 320 input should be smaller. The change of anchor size could gain performance improvement.
3. Evaluation on https://github.com/Adamdad/Object-Detection-Metrics.git
4. I only use the pure model of YOLOv3-Mobilenet with no additional tricks.

Guide of keras-yolov3-Mobilenet

1.train_Mobilenet.py

• Code for training
• I change some of the code to read in the annotaions seperately (train.txt and val.txt), remember to change that, and the .txt file are in the same form descibed below

2.yolo3/model_Mobilenet.py

• Model_Mobilenet is the yolo model based on Mobilenet
• If you want to go through the source code,ignore the other function,please see the yolo_body (I extract three layers from the Mobilenet to make the prediction)

3.yolo_Mobilenet.py

• Testing on images

Be sure that you do not load pretrained model when training because I did it on keras_applications,and the keras will load the pretrained model for you

if you find anything tricky, contact me as you wish

---

Evaluation

Please use this repo to draw the RP curve calculate the MAP https://github.com/Adamdad/Object-Detection-Metrics.git

---

Guide of keras-yolov3

[this is the guide for darknet53 not mobilenet]

Quick Start

1. Download YOLOv3 weights from YOLO website.
2. Convert the Darknet YOLO model to a Keras model.
3. Run YOLO detection.

wget https://pjreddie.com/media/files/yolov3.weights
python convert.py yolov3.cfg yolov3.weights model_data/yolo.h5
python yolo_video.py [OPTIONS...] --image, for image detection mode, OR
python yolo_video.py [video_path] [output_path (optional)]

For Tiny YOLOv3, just do in a similar way, just specify model path and anchor path with --model model_file and --anchors anchor_file.

Usage

Use --help to see usage of yolo_video.py:

usage: yolo_video.py [-h] [--model MODEL] [--anchors ANCHORS]
                     [--classes CLASSES] [--gpu_num GPU_NUM] [--image]
                     [--input] [--output]

positional arguments:
  --input        Video input path
  --output       Video output path

optional arguments:
  -h, --help         show this help message and exit
  --model MODEL      path to model weight file, default model_data/yolo.h5
  --anchors ANCHORS  path to anchor definitions, default
                     model_data/yolo_anchors.txt
  --classes CLASSES  path to class definitions, default
                     model_data/coco_classes.txt
  --gpu_num GPU_NUM  Number of GPU to use, default 1
  --image            Image detection mode, will ignore all positional arguments

---

4. MultiGPU usage: use --gpu_num N to use N GPUs. It is passed to the Keras multi_gpu_model().

Training

1. Generate your own annotation file and class names file.
   One row for one image;
   Row format: image_file_path box1 box2 ... boxN;
   Box format: x_min,y_min,x_max,y_max,class_id (no space).
   For VOC dataset, try python voc_annotation.py
   Here is an example:

   path/to/img1.jpg 50,100,150,200,0 30,50,200,120,3
   path/to/img2.jpg 120,300,250,600,2
   ...
   

2. Make sure you have run python convert.py -w yolov3.cfg yolov3.weights model_data/yolo_weights.h5
   The file model_data/yolo_weights.h5 is used to load pretrained weights.

3. Modify train.py and start training.
   python train.py
   Use your trained weights or checkpoint weights with command line option --model model_file when using yolo_video.py Remember to modify class path or anchor path, with --classes class_file and --anchors anchor_file.

If you want to use original pretrained weights for YOLOv3:
1. wget https://pjreddie.com/media/files/darknet53.conv.74
2. rename it as darknet53.weights
3. python convert.py -w darknet53.cfg darknet53.weights model_data/darknet53_weights.h5
4. use model_data/darknet53_weights.h5 in train.py

---

Some issues to know

1. The test environment is

   • Python 3.5.2
   • Keras 2.1.5
   • tensorflow 1.6.0

2. Default anchors are used. If you use your own anchors, probably some changes are needed.

3. The inference result is not totally the same as Darknet but the difference is small.

4. The speed is slower than Darknet. Replacing PIL with opencv may help a little.

5. Always load pretrained weights and freeze layers in the first stage of training. Or try Darknet training. It's OK if there is a mismatch warning.

6. The training strategy is for reference only. Adjust it according to your dataset and your goal. And add further strategy if needed.

7. For speeding up the training process with frozen layers train_bottleneck.py can be used. It will compute the bottleneck features of the frozen model first and then only trains the last layers. This makes training on CPU possible in a reasonable time. See this for more information on bottleneck features.

Citation

Please cite MobileNet-YOLO in your publications if it helps your research:

@article{MobileNet-Yolov3,
     Author = {Adam Yang},
     Year = {2018}
}
 @article{yolov3,
     title={YOLOv3: An Incremental Improvement},
     author={Redmon, Joseph and Farhadi, Ali},
     journal = {arXiv},
     year={2018}
}
@article{mobilenets,
     title={MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications},
     author={Andrew G. Howard, Menglong Zhu, Bo Chen,Dmitry Kalenichenko,Weijun Wang, Tobias Weyand,Marco Andreetto, Hartwig Adam},
     journal = {arXiv},
     year = {2017}
}