Synthetic Bounding-Box Regression Database Tool

A Python tool to create synthetic bounding-box data for evaluation of deep regression techniques.

Getting Started

• Clone the repo into your folder

  git clone https://github.com/sanjeev309/synthetic_bbox_regression_db_tool.git

• Open main.py and change values of TOTAL_IMAGES, IMAGE_DIMENSION and PADDING if required.

• Run the program using python3 main.py

• The Images and target co-ordinates are saved in numpy format in npy files data.npy and target.npy respectively.

• The images are stacked along the 0th axis in data whereas the corresponding index of target contains the 4 coordinates [x1, y1, x2, y2]

  • where x1,y1 are co-ordinates of top-left corner of rectange res.
  • where x2,y2 are co-ordinates of bottom-right corner of rectange res.

Prerequisites

Python3 and OpenCV 3.2

Data Structure

data.npy

target.npy

Sample Data

Data :

Target :

[0.37, 0.31, 0.91, 0.81]

Author

• Sanjeev Tripathi - LinkedIn

License

This project is licensed under the MIT License - see the LICENSE file for details