sefaburakokcu / finn-quantized-yolo

Low-Precision YOLO on PYNQ with FINN
26 stars 6 forks source link

Add different types of object detection #6

Open ttm17 opened 1 year ago

ttm17 commented 1 year ago

Hi, I have followed the steps at here https://medium.com/@bestamigunay1/end-to-end-deployment-of-lpyolo-low-precision-yolo-for-face-detection-on-fpga-13c3284ed14b to obtain the .bit and .hwh file for yolov3-tiny.

How can I modify it so that I can detect other objects in COCO dataset other than only detect faces? I get this error when I modify the nc = 80 RuntimeError: shape '[1, 3, 85, 13, 13]' is invalid for input of size 3042

JPPalacios commented 1 month ago

@ttm17 Hi, were you able to deploy your COCO-based multi-class detector onto hardware? I have several multi-class models deployed with poor/random predictions. Did you modify the common.py file at all?