-
Notifications
You must be signed in to change notification settings - Fork 0
/
README
17 lines (12 loc) · 1.02 KB
/
README
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
PROJECT NAME : poster_detection_homography_and_template_matching
LAST UPDATE : 21/06/21
MAIN TASK : 1. Detect (classify and localize) templates (posters) on realograms (real scene)
using keypoints+homography (poster_detection.py)
2. in case the template doesnt have enough valuable key points, homography will fail to match between the template
and it place in the scene , so we decided to try template matching algorithm on different scales ,
and by analyzing correlation map scores for each scale (5 per template) and each template class (currently only 4 'hard' templates)
(scaled-template-matching.py)
METHODS USED IN THIS PROJECT: key points extractors , homography , template matching , region props , binary image segmentation ,
region charcaristics (bbox area , region area , number of points , etc)
ASSETS: - short presentation created for weekly CV meeting
- jupyter notebook with couple of experiments