An implement of origin faster-rcnn based on tensorflow 2.x and visualize it.
使用Tensorflow2.x实现了faster-rcnn并可视化算法运行流程
最近忙着写毕业论文，issue里提到的问题后面都会跟进的。包括 训练过程不稳定、用vgg16预训练权重、predict函数没有写完整等等，写完论文会改的 =。=
https://zhuanlan.zhihu.com/p/157907380
https://blog.csdn.net/cmd233333/article/details/107170022
both for Chinese readers.
but the image is easy to understand and I believe English readers could see it.
any question, just issue or contact me at cmd2333@qq.com
tensorflow 2.x
tensorboard
改变config.voc_data_dir为本地VOC数据集路径，运行trainer.py即可。
VOC2007(or any customed dataset in the form of VOC)
you could download it by its official website or Baidu CouldDisk as below
link: https://pan.baidu.com/s/170jE4b1t841g1Ax_2i9P6g
password: uxg8