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.
-
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