Tianhan Hu:
This week, I continued working on hand gesture detection.
Our current implementation on hand gesture detection relies on extracting contour of hand on its first step. To achieve that, we are using thresholding with fixed values.
As pointed out by Professor Marios in the design review presentation, thresholding with fixed values is troublesome in that people could differ in skin colors, and is largely dependent on the environment such as the lighting condition. Thus, this week I am primary researching possible ways of achieving adaptive threshold.
Currently, I am still reading papers on IEEE, and have not started implementing my own algorithm. I expect to start implementation next week, and make major progress in spring break.
Xinyu Zhao:
This week I kept improving the data extraction from images. I was also thinking about alternatives of detection algorithms and learned about the yolo object detector with tutorial. I’m waiting for the ordered cameras which have different specs to actually have more meaningful first-hand tests on the effects. I think I’m behind schedule and next week I’ll definitely spend more time on it and make sure to test the camera when received.
Next week and during the spring break I’ll think more about the how adjustment of the camera angle affects the detection after trying the cameras, and provide a report on the findings.
0 Comments