GitHub - landongw/card_recognition: Project to find specific playing cards within a deck.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 18 Commits
.vscode		.vscode
ENV		ENV
__pycache__		__pycache__
pyimagesearch		pyimagesearch
training		training
.gitignore		.gitignore
README.txt		README.txt
RGB_slider.py		RGB_slider.py
SIFT_example.py		SIFT_example.py
SURF_example.py		SURF_example.py
canny_edge.py		canny_edge.py
contour.py		contour.py
corner_detection.py		corner_detection.py
detect_shapes.py		detect_shapes.py
fast_corner.py		fast_corner.py
haar_cascade_detection.py		haar_cascade_detection.py
main.py		main.py
match_rectangle.py		match_rectangle.py
shi_tomasi_corner.py		shi_tomasi_corner.py
sift_keypoints.jpg		sift_keypoints.jpg
surf_keypoints.jpg		surf_keypoints.jpg
template_match.py		template_match.py
train_find_image.py		train_find_image.py

Repository files navigation

THIS APPLICATION IS AN EXPERIMENT WITH OPENCV


USAGE:

VEC GENERATION:

opencv_createsamples \
-vec /Users/landonwiedenman/Documents/lc101/hacker-problems/card_recognition/training/positive.vec \
-img /Users/landonwiedenman/Documents/lc101/hacker-problems/card_recognition/images/sevenhearts.png \
-bg /Users/landonwiedenman/Documents/lc101/hacker-problems/card_recognition/training/bg.txt \
-maxxangle 0.1 \
-maxyangle 0.1 \
-maxzangle 0.1 \
-w 20 \
-h 30




CASCADE TRAINING:

opencv_traincascade \
-data /Users/landonwiedenman/Documents/lc101/hacker-problems/card_recognition/training/cascade \
-vec /Users/landonwiedenman/Documents/lc101/hacker-problems/card_recognition/training/positive.vec \
-bg /Users/landonwiedenman/Documents/lc101/hacker-problems/card_recognition/training/bg2.txt \
-numPos 40 \
-numNeg 40 \
-numStages 7 \
-featureType HAAR \
-minHitRate 1 \
-maxFalseAlarmRate 0.4 \
-w 20 \
-h 30





TROUBLSHOOTING DOCS:

https://stackoverflow.com/questions/16058080/how-to-train-cascade-properly
http://answers.opencv.org/question/10872/cascade-training-error-opencv-244-train-dataset-for-temp-stage-can-not-filled-branch-training-terminated-cascade-classifier-cant-be-trained-check-the/