li-s / Object-detection

Detecting the positions of objects in images

Object-localization

Detecting the positions of objects in images For introduction, explanation, and readings on object detection, go to Image-recognition repository.

Introduction

Getting started

Prerequisites

• Python3
• tensorflow
• numpy
• scikit-image
• matplotlib

Usage

1. Run get_predictions.py to get the predictions of label, confidence, and boudning box saved in /cache/predictions.pkl.
2. Run make_detfile.py to create detection files of the images under ../data/predictions/.
3. Run py_faster_rcnn_evaluate.py to get the average precision of each class and mean average precision.

or

1. Run predict.py [image_path] to find the prediction, confidence and bounding box of one image.

API reference

• ssd keras

Miscellaneous

Readings

• Single shot multibox detector research paper