Closed WangzcBruce closed 11 months ago
I am sorry to bother you. I ran the visual.py, then only got the predictions. How can i get the gt?
Hi,
Since I only visualized GT when I generated the demo, I didn't include it in my code. But all it takes to accomplish this function is a simple modification to visual.py
:
Instead of loading the input images for inference, directly load the GT from the dataloader. (Line 113-122)
For a simple implementation, you can assign labels['segmentation']
directly to output['segmentation']
and assign labels['flow']
directly to output['instance_flow']
.
Sorry I didn't keep my original implementation. If you still have difficulity in it, feel free to raise your question further.
There was a little trouble to replace the predictions with inputs simply. The dims between them are different.
This function affected the operation of replacing the preditions with gts.
There was a little trouble to replace the predictions with inputs simply. The dims between them are different.
Hi,
You are right. While the previously mentioned idea is correct, modifying the visualization really isn't that simple.
For your convenience, I've updated a version V1.1 and added the GT visualization to the original code.
You just need to follow the tutorial and add VISUALIZATION.VIS_GT True
after the original command.
The above modifications work fine on my computer. Please contact me further if you encounter any more problems.
I am sorry to bother you. I ran the visual.py, then only got the predictions. How can i get the gt?