Team Status Report for 04/16

For this week, the biggest risk is still OpenCV. The current algorithm doesn’t have sufficient accuracy possibly due to low resolution camera. After talking with Prof. Kim, we temporarily gave up on including car color as an additional feature to recognize cars. Instead, we’ve tried multiple preprocessing techniques that may enhance image quality before feeding into the model (detail in Ke’s report). Currently, the most feasible plan is to send images to the server and process plate recognition with more advanced model on the server side. Meanwhile, we’re also working on the navigation algorithm as advised by Prof. Kim. The change of plan mostly comes from the limitation of OpenCV algorithm, and server side also needs to be modified to accommodate the image processing component.