yrong / flowchart-detection Goto Github PK

View Code? Open in Web Editor NEW

Detecting hand drawn flowcharts using Tensorflow Object Detection API - Faster RCNN

Python 48.07% Shell 0.68% GLSL 0.01% C++ 43.31% Jupyter Notebook 4.10% Makefile 0.06% Dockerfile 0.02% JavaScript 0.05% HTML 0.23% TypeScript 0.01% C 0.30% LLVM 0.01% CMake 0.97% Java 0.59% Objective-C 0.01% Objective-C++ 0.10% Ruby 0.01% Go 1.46% Perl 0.01% PureBasic 0.04%

flowchart-detection's Introduction

Flowchart-Detection

Detecting hand drawn flowcharts using Tensorflow Object Detection API

This repository uses Tensorflow Object Detection API for detecting hand drawn flowcharts. The dataset images are located in the folder models/research/object_detection/images/test

I have used the Faster-RCNN Inceptionv2 model for detecting the flowcharts.

Output

Running Locally

Install the Tensorflow Object Detection API

#From Flowchart-Detection/models
export PYTHONPATH=$PYTHONPATH:`pwd`:`pwd`/research:`pwd`/research/slim:`pwd`/research/object_detection

#From Flowchart-Detection/
python Object_detection_image.py        (#You can set any input image of your choice located in models/research/object_detection/images/test inside this script)

Coordinates of the bounding boxes will be stored in coordinates.json

NOTE: OpenCV is needed for displaying the image. It can be installed using pip install opencv-python

Recommend Projects

yrong / flowchart-detection Goto Github PK