penny4860/retinanet-digit-detector

Implemented digit detector in natural scene. I used SVHN as the training set, and implemented it using keras-retinanet.

Retinanet Digit Detector

I have implemented retinanet digit detector using keras-retinanet package and svhn dataset.

Usage for python code

0. Requirement

• python 3.6
• tensorflow 1.10.0
• keras 2.2.4
• keras-retinanet 0.5.0

1. Digit Detection using pretrained weight file

In this project, the pretrained weight file is stored in resnet50_full.h5. Test set evaluation (13068-images) score is mAP: 0.8148

• Download resnet50_full.h5 to the project-root/snapshots
• Run infer.py

You can see the following figure:

2. SVHN dataset in Pascal Voc annotation format

In this project, I use pascal voc format as annotation information to train object detector. An annotation file of this format can be downloaded from svhn-voc-annotation-format.

3. Training from scratch

Please refer to main.py