zcdliuwei / TensorFlow2.0_SSD

A tensorflow_2.0 implementation of SSD (Single Shot MultiBox Detector) .

TensorFlow2.0_SSD

A tensorflow_2.0 implementation of SSD (Single Shot MultiBox Detector) .

Requirements:

• Python >= 3.6
• TensorFlow >= 2.0.0-rc1
• Pillow >= 6.1.0

Usage

Train on PASCAL VOC 2012

1. Download the PASCAL VOC 2012 dataset.
2. Unzip the file and place it in the 'dataset' folder, make sure the directory is like this :

|——dataset
    |——VOCdevkit
        |——VOC2012
            |——Annotations
            |——ImageSets
            |——JPEGImages
            |——SegmentationClass
            |——SegmentationObject

3. Run train.py to start training, before that, you can change the value of the parameters in configuration.py.

References

• The paper: SSD: Single Shot MultiBox Detector
• http://zh.gluon.ai/chapter_computer-vision/ssd.html
• https://blog.csdn.net/XiangJiaoJun_/article/details/84503224
• MXNet tutorial on bilibili :
  (1) https://www.bilibili.com/video/av16012497
  (2) https://www.bilibili.com/video/av16225415
  (3) https://www.bilibili.com/video/av16440968