-
Notifications
You must be signed in to change notification settings - Fork 5
Generation Code Explanation
Christopher Rodriguez edited this page Jul 25, 2017
·
1 revision
The droplet generation script is fairly straightforward.
- Split the target video into frames and save these images into the frames directory.
- For each frame, crop the image to the user defined detection box.
- Blur the image.
- Apply Canny's edge detection.
- Get the average pixel value in the frame.
- Once all frames are processed, smooth the resulting data using the input weights.
- Find the local maxes of this data set.
- For each local max, get the detection box at that frame.
- Find the convex hull area of the detected contours of the frame.
- Save this area as the droplet area in pixels.
- The drop generation rate is the number of local maxes divided by the number of frames.
- Convert the drop areas by multiplying by the conversion factor (user drawn distances divided by user drawn pixel distance) squared.
- Convert the drop generation by multiplying by FPS