Open juniel03 opened 5 years ago
Hey. Thanks :) Can you post your error and config file? I think conversion it is possible with a right training. Without see error, first idea is that you have to check fame size. It must be equal to [288, 608] with all possible size between adding 32 (288, 320, 352, 384, 416...) both in the top of cfg file and in CameraActivity
Size frame_size = new Size(416, 416);
Mat blob = Dnn.blobFromImage(frame, 1.0 / 255.0, frame_size, mean, true, false);
when i trained my own yolov3 readdnnfromdarknet gives null into net variable. do you have solution for this then please share how to load custom trained model?
hey there great application thanks for the idea! i was wondering if it can be converted to face detection? i tried to change the weight and config files but I just got errors