Steps to Create Vision based model using Custom Vision for object detection
Assuming video is available for with objects.
Steps to Follow:
- Annotate images with bounding boxes tools like Via
- Convert VOC to pascal format
- Use Notebook in Azure ML to format the data to custom vision
- Log into Custom Vision validate the bounding boxes
- Train the model
- Test - Validate model
VIA tool links
http://www.robots.ox.ac.uk/~vgg/software/via/
Online Tool:
http://www.robots.ox.ac.uk/~vgg/software/via/via.html
Code for Python based upload images and bounding box into azure custom vision.