YOLO-Vehicle-Detection

Vehicle detection using YOLOv8 pretrained model on custom dataset.

Classification : Labelling of an object with no bounding box

Localization : Classification of an object with bounding box

Detection : Locallizing multiple objects

In our model, we bound image with labels and send it in a convolutional net and softmax will give us the classification.

%y = pc does image have any object/ objectness score, % bx bounding box cordinates, % by
% bz % bw % c1 probability of class 1, % c2 probability of class 2, % c3 probaiblity of class 3,

loss = (y1-yp1)^2 + (y2-yp2)^2 + ...

We do this in our feature map by Conv