understanding pytorch pycoco tools object detection output

Question

I am using pytorch vision library for object detection. I am using utilities provided for objection detection metrics. https://github.com/pytorch/vision. I am seeing following output

Epoch: [2] [0/10] eta: 0:00:50 lr: 0.000500 loss: 1.1589 (1.1589) loss_classifier: 0.1807 (0.1807) loss_box_reg: 0.0592 (0.0592) loss_objectness: 0.6662 (0.6662) loss_rpn_box_reg: 0.2528 (0.2528) time: 5.0786 data: 1.0029 max mem: 8476

My question is what is numbers in the brackets denote? Another quesiton is what is difference between and loss and loss_classifier in this context. Kindly help me to understand output. Thanks for your time and help.

Answer 1

Looking at the source code, it seems that the first value is the median value for the epoch and the second value in parentheses is the global average/mean. The difference between the value for loss and loss_classifier is that the value for loss is the sum of the losses of the individual parts (including that of the classifier, 0.1807 + 0.0592 + 0.6662 + 0.2528 = 1.1589), whereas the value for loss_classifier is the loss for just the classifier of the model, i.e. the part that classifies the pixels into the different classes.