Every model is implemented in only one file! It is easy for people to understand!

Yolo2

Yolo3

SSD

RetinaNet

RefineDet

Light Head Rcnn

PFPNet

CenterNet

FCOS

1. voc format dataset

2. fill in utils.voc_classname_encoder.py

3. run utils.test_voc_utils.py

fill in dict 'image_augmentor_config' in test-model.py

see utils.image_augmentor.py for details

see https://github.com/Stick-To/Online_Image_Augmentor_tensorflow for details

fill in dict 'config' in test-model.py

run test-model.py

The pre-trained vgg_16.ckpt could be downloaded from http://download.tensorflow.org/models/vgg_16_2016_08_28.tar.gz

run annotated code in test-model.py

see utils.tfrecord_imagenet_utils.py

https://github.com/Stick-To/Deep_Conv_Backone

corresponding repository in https://github.com/Stick-To

python3.7 tensorflow1.13