I have trained the model with 2 class of images [ Mask_detected and No_Mask detected] ,Where I ran my model in Google Colab till 1300 iteration which gives me the Highest Accuracy when training with 1113 images .
Upload the below files to darknet/data directory : For training
train.txt obj.names obj/data dataset test.txt ( optional )
After training use the best weight file for detection .
Further Resource : https://github.com/AlexeyAB/darknet
Model Ouput :
Mask_Not_Detected
Model_Summary
class_id = 0, name = Mask_Detected (TP = 681, FP = 29)
class_id = 1, name = Mask_Not_Detected (TP = 385, FP = 18)
precision = 0.96, recall = 0.96, F1-score = 0.96
TP = 1066, FP = 47, FN = 40, average IoU = 71.77 %
mean average precision (mAP@0.50) = 0.973706, or 97.37 % Total Detection Time: 33 Seconds
If you have doubts please inbox me at edugan28@gmail.com