My current classification model accurately classifies 25 native and general classes on the street. But the problem comes when classifying people and people on rickshaws, bikes, etc. All classified as person. I trained an SVM for further classification, but the testing result wasn’t good enough. I can train a neural network, but first, I was wondering if there is another way to solve this problem—a mathematical solution or a lightweight context-based solution. (rickshaw is a tri-cycle like vehicle)
