Use DTW algorithm to match the optimal path of audio and music scores, and backtrack the path to calculate feature scores such as pitch, rhythm completeness, etc.
Process audio data, perform calibration training on each feature score through multiple linear regression, and get a scoring model.
Build an SSD network for target detection, detect and classify the power bank in the subway security scan map (whether it has a chip or not)
Expand the unbalanced samples through rotation, cropping and other methods, and design a weighted loss function
Cloud Print of Nankai University
- Description: It's an open source project which lead by several university students. The projects intends to provide a platform for university students to print their materials conveniently.
Developed by PHP with ThinkPHP.
I developed the background management of the system.