Save loss chart in google Colab without OpenCV

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )

http://pjreddie.com/darknet/

Other

21.76k stars 7.96k forks source link

Save loss chart in google Colab without OpenCV #7228

Open arczer9 opened 3 years ago

arczer9 commented 3 years ago

I have build darknet in google Colab without OpenCV and now i'm training my detector with this command: !./darknet detector train obj/person.data obj/yolov3-tiny.cfg obj/darknet.conv -dont_show -mjpeg_port 8090 -map 0.5

Will this save chart of Loss and mAP as chart.png after training is finished?

AlexeyAB commented 3 years ago

OpenCV is required to build Loss and mAP chart.
-mjpeg_port 8090 - it will not work in Google Colab, because Colab blocks all outgoing connections
If you will compile Darknet with OpenCV, the Loss and mAP chart will be saved to chart.png file, that you can download from Colab