[mmclassification 1.X]How can I record the testing inference results of each images in each category #1444
Unanswered
ThomaswellY
asked this question in
Q&A
Replies: 2 comments 4 replies
-
Use |
Beta Was this translation helpful? Give feedback.
4 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
following the mmclassification 1.X tutor manual,the testing inference is successful, and the record on log is shown as follow: (max_epoch=1, logger_inter=100)
{"lr": 0.1, "data_time": 0.0008228302001953125, "loss": 1.4076386213302612, "time": 0.08029277324676513, "epoch": 1, "memory": 2455, "step": 100}
{"lr": 0.1, "data_time": 0.0007920742034912109, "loss": 1.3337437868118287, "time": 0.07977149486541749, "epoch": 1, "memory": 2455, "step": 200}
{"lr": 0.1, "data_time": 0.0008879423141479493, "loss": 1.3094250798225402, "time": 0.08056514263153076, "epoch": 1, "memory": 2455, "step": 300}
{"single-label/precision": 42.19956970214844, "single-label/recall": 35.5002326965332, "single-label/f1-score": 29.484556198120117, "accuracy/top1": 46.23514175415039, "data_time": 0.00510249137878418, "time": 0.02877798080444336, "step": 1}
However, the final evaluation of testing inferece is over the whole testing dataset, which is not enough for me.
I have consult "mmengine.readthedoc.io" for help, and seems mmengine.hook or mmengine.runner doesn't support such functional requirements.
by the way , in branch of mmclassification 0.X, the whole testing inference result is in form of vector , and can be ultilized easily from "outputs = single_gpu_test(model, data_loader, args.show, args.show_dir,**show_kwargs)"in tools/test.py. while in 1.X branch, the corn is hidden deeper.
Sincerely looking forward to your response, Thanks in advance~
Beta Was this translation helpful? Give feedback.
All reactions