Learning OpenCV 4 Computer Vision with Python 3 - Third Edition

Combining an SVM with a sliding window

By combining our SVM classifier with a sliding window technique and an image pyramid, we can achieve the following improvements:

• Detect multiple objects of the same kind in an image.
• Determine the position and size of each detected object in an image.

We will adopt the following approach:

1. Take a region of the image, classify it, and then move this window to the right by a predefined step size. When we reach the rightmost end of the image, reset the x coordinate to 0, move down a step, and repeat the entire process.
2. At each step, perform a classification with the SVM that was trained with BoW.
3. Keep track of all the windows that are positive detections, according to the SVM.
4. After classifying every ...