Question/Issue:
Is Object Recognition training suitable for pathfinding applications?
Project ID:
davefer-pathfinder
Context/Use case:
This is my first attempt at a computer vision model. I’m trying to train a model that will identify suitable paths for my rover. I have 671 photos that I harvested from a video I made.
So far I’ve worked through almost 500 of these photos labeling (with rectangles) “Path” and “Obstacle” items.
Is this proper basic methodology for training a pathfinding model? Or is there some other method that allows me to draw a line in an image where I see a suitable path?
Steps Taken:
- Upload hundreds of photos (of walking route: sidewalks, dirt paths, etc)
- [ongoing] Train model by using rectangles to label either “Path” or “Obstacle”
Expected Outcome:
Pathfinding model for visual navigation of rover on ground.