Final project for (Computer) Vision and Cognitive System

• numpy
• opencv-python

For each video frame:

Predict a ROI for each painting:

1. CLAHE (Contrast Limited Adaptive Histogram Equalization)
2. Edge Detection with Sobel
3. Bilateral Filtering
4. Thresholding
5. Morphology Transformations
6. Significant Contours (cv2.findContours)
7. Contours refining:
   • Find Bounding Boxes (cv2.boundingRect)
   • Merge overlapping
   • Convex hull
8. Discard false positives:
   • Check dimensions and aspect ration
   • Histogram distance & update

Starting from contours found in previous point and considering one contour at a time:

1. Polygonal approximation (cv2.approxPolyDP)
2. Find lines with Hough transform
3. Compute lines intersections
4. Average vertices with K-Means
5. Order vertices
6. Compute aspect-ratio
7. Warp perspective

Match each detected painting to the paintings DB:

1. Find descriptors with ORB
2. Find best matches (BFMatcher with Hamming normalization)
3. Find room in which paintings are collocated

Predict a ROI around each person:

1. YOLO v3 (from OpenCV)

   For each detection:

   • Predict a score for each class
   • Take only the score belonging to the person class
   • Thresholding
   • Non-maximum suppression

2. Discard people in paintings