Qt program for reading & recording device
OpenCV build with contrib:https://jbox.sjtu.edu.cn/l/FoibN4
Detect the point of the pen using Hough
Replace the ROI by Maching Learning (using python)
Convert the position on the projection to the point on the book
Train the model so that the loss decrease below 1 stably
Debug & calibrate the projection area and tracking area
Some Calibration UI
Config the tensor flow environment, train the model: