open-edge-platform
Where can I save the segmentation results separately
sorry to bother you ! Where can I save the segmentation results separately？Because I want to get the specific coordinates of the defect. And I can't find the presion and recall, I also want to output them.
@wangk98, PR #298 would potentially addresses your issue. Once merged, we could add some documentation
Regarding setting different metrics, you could modify the configuration as follows:
@wangk98, regarding the segmenttaion results, do you want the segmentation mask to be saved separately? Or would like to have access to the annotations from the inferencer?
@samet-akcay when running python tools/inference.py we can get the result that looks like
(1st image)
But we would want to have an option for getting the segmentation result also (without overlay), it looks like below:
(2nd image)
There is an argument: "--overlay_mask" in inference.py but when I set it to True then it will be saved as a result with both overlay and segmentation.
(3rd image)
We would also get segmentation results (like as 2nd image) when running inference. Could you tell us how to do it?
@wangk98, regarding the segmenttaion results, do you want the segmentation mask to be saved separately? Or would like to have access to the annotations from the inferencer?
yes, I want the segmentation mask to be saved separately. If possible, I would also like to be able to plot the PR curve and AUC curve directly
yes, I want the segmentation mask to be saved separately. If possible, I would also like to be able to plot the PR curve and AUC curve directly
You could follow PR #429 for metric visualization
@nguyenanhtuan1008, have you tried the new inference? If you use tools/inference/lightning_inference.py, you could get that output. For instance, this is what I get when I use the new lightning inference
Closing this due to inactivity. Fee free to re-open if you still encounter any issues.
For me the tools/inference/lightning_inference.py won't work with cflow model
I am getting this error, while running lightning_inference.py with cflow model in superimpose_anomaly_map superimposed_map = cv2.addWeighted(anomaly_map, alpha, image, (1 - alpha), gamma) cv2.error: OpenCV(4.6.0) /io/opencv/modules/core/src/arithm.cpp:647: error: (-209:Sizes of input arguments do not match) The operation is neither 'array op array' (where arrays have the same size and the same number of channels), nor 'array op scalar', nor 'scalar op array' in function 'arithm_op'
Predicting DataLoader 0: 0%| | 0/1 [00:00<?, ?it/s]
I've just checked. This issue is related to #568. You could follow the progress there.