From our understanding of CNN and image classification so far, let's try to understand how we can approach the object detection problem, and that should logically lead us to the discovery of the underlying complexity and challenges. Assume we are dealing with monochromatic images for simplicity.
Any object detection at a high level may be considered a combination of two tasks (we will refute this later):
- Getting the right bounding boxes (or as many of them to filter later)
- Classifying the object in that bounding box (while returning the classification effectiveness for filtering)
So, object detection not only has to cater to all the challenges of image classification ...
