mmsegmentation
mmsegmentation copied to clipboard
open-mmlab
OpenMMLab Semantic Segmentation Toolbox and Benchmark.
I use upernet+internimage model of mmseg=0.30.0 to train segmentation data, and mmcv version is 1.7.2, So how can change code to run it?
Traceback (most recent call last): File "tools/train.py", line 245, in main() File "tools/train.py", line 234, in main train_segmentor( File "/home/a8/zy/Projects/Semantic Segmentation/SCTNet/mmseg/apis/train.py", line 194, in train_segmentor runner.run(data_loaders, cfg.workflow) File "/home/a8/anaconda3/envs/STCNet/lib/python3.8/site-packages/mmcv/runner/iter_based_runner.py", line...
The download link for the ADE20K dataset in the dataset_prepare.md file doesn't seem to work, it can't be downloaded safely and why is the conversion file for the ADE20K dataset...
when we consider resize process and size divisibility in data preperation , your pipeline seems likie original image -> resize image -> pad img which refering to following shape record...
Hello, I encountered the following problem when using benchmark.py to conduct FPS testing on my own data set. Can you give me some guidance? Thank you! My mmseg version is...
Shouldn't the `reduce_zero_label=True` be set in the COCO-Stuff 164k base config? The docstring for the COCOStuffDataset also mentions that it should. The index zero is for the "unlabeled" class.
## Motivation I was working through using mmsegmentation and some of the steps did not work. Trial and error led me to being able to use it so I thought...
Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand...
Hello, I am just wondering how to get validation loss in the current version where previous issues mentioned val_step or workflow. But all of those do not work now, so...
Metadata
Owner
Metadata
OpenMMLab Semantic Segmentation Toolbox and Benchmark.