Existing models on images of cars taken horizontally? [closed]
Hi there
I´m researching on how to find the best approach on this project of mine. I do a lot of tasks counting vehicles in stitched images taken from my drone (camera pointed towards ground level). So this is pictures of cars seen from above, where there' s a rich access of pictures of cars taken from street level on the web. I´m unsure if there are existing models or similar (casacades, YOLO, coco-datasets etc) that can get me started and avoiding training my own model on cars seen from UAV.
What should I look after? Thank you.
I don't understand what's the link between the OpenCV library and a DNN dataset...????
There was a project to detect trucks at night but it was not free. They would steady the drone at a consistent location and process movement. Sadly, it was abandoned due to opinionated bickering. Otherwise perfect.
You should do do a dataset search for uav + cars. Then you can train on it. If you are lucky theres a pretrained model for it, google for this too.