mobilenetv2-yolov3

use mobilenetv2 replace darknet53 for yolov3 detection

Comment

Mobilenetv2-Yolov3

Tensorflow implementation mobilenetv2-yolov3 inspired by keras-yolo3

Update

Backend:

  • [x] MobilenetV2
  • [x] Efficientnet
  • [x] Darknet53

Callback:

  • [x] mAP
  • [ ] Tensorboard extern callback

Loss:

  • [x] MSE
  • [x] GIOU

Train:

  • [x] Multi scale image size
  • [x] Cosine learning rate

Tensorflow:

  • [x] Tensorflow2 Ready
  • [x] Use tf.data to load dataset
  • [x] Use tfds to load dataset
  • [x] Remove image shape input when use session
  • [ ] Convert model to tensorflow lite model
  • [x] Multi GPU training
  • [ ] TPU support

Serving:

  • [x] Tensorflow Serving warm up request
  • [x] Tensorflow Serving JAVA Client
  • [x] Tensorflow Serving Python Client
  • [x] Tensorflow Serving Service Control Client

- [ ] Tensorflow Serving Server Build and Plugins develop

Usage

Install:

pip install -r requirements.txt

Get help info:

python main.py --help

Train:

First: Format file name like [name]_[number].[extension]
Second: If you are using txt dataset, please format records like [image_path] [,[xmin ymin xmax ymax class]]
(for convenience, you can modify voc_text.py to parse your data to specific data format), else you should use modify voc_annotation.py, then run 

python voc_annotation.py

to parse your data to tfrecords.
Third: Run 

python main.py --mode=TRAIN --train_dataset_glob=<your dataset glob>

Predict:

python main.py --mode=IMAGE --model=<your_model_path>

Export serving model:

python main.py --mode=SERVING --model=<your_model_path>

Use custom config file:

python main.py --config=mobilenetv2.yaml

Resources

  • Download pascal tfrecords from here.
  • Download pre-trained mobilenetv2-yolov3 model(VOC2007) here

Performance

Network: Mobilenetv2+Yolov3
Input size: 416*416
Dataset: VOC2007
mAP: 

aeroplane ap:  0.585270123970341
bicycle ap:  0.7311717479746895
bird ap:  0.6228634475289679
boat ap:  0.44729361226611786
bottle ap:  0.3524265151288485
bus ap:  0.7260233058709467
car ap:  0.7572503412774444
cat ap:  0.8443930169586521
chair ap:  0.3530240979604032
cow ap:  0.5680746465428056
diningtable ap:  0.6046673143934721
dog ap:  0.8096497542858805
horse ap:  0.785358647511358
motorbike ap:  0.7299038925396009
person ap:  0.6926967393665762
pottedplant ap:  0.2960290730045794
sheep ap:  0.5569735405574012
sofa ap:  0.6053534702293342
train ap:  0.7304618425853895
tvmonitor ap:  0.5983913977616169
mAP:  0.6198638263857212

GPU inference time (Python+MX150): 78ms
CPU inference time (Python+i7-8550U): 112ms
Model size: 37M

Reference

paper:

Built With

Share this project:

Updates