semantic-segmentation-pytorch

semantic-segmentation-pytorch copied to clipboard

Training on ADE20K

5

Hello! I try to train default models from train.py on ADE20K, but I see such error: ``` Input arguments: weights_decoder batch_size_per_gpu 2 weights_encoder workers 16 lr_pow 0.9 seed 304 epoch_iters...

loveis98

When I use ```pytorch 1.3.1``` with ```python3.7.4```, like this ``` import torch from torch.utils.data.dataloader import _DataLoaderIter ``` Here is an errer : ```cannot import name '_DataLoaderIter' from 'torch.utils.data.dataloader'``` How should...

MuQQQQQ

Dear authors: Thanks for your great work. COCO is widely used, hence i wonder will you add coco panoptic dataset support? or can you give me some tips on how...

dragen1860

Downsampling segmentation labels vs Upsampling predictions

2

Hi guys, Thanks for having this awesome repo. I'm writing some code for semantic segmentation and noted that you are downsampling labels instead of upsampling predictions. This issue is not...

aicaffeinelife

I get the following error when I run inference: ``` [2019-10-30 15:33:25,356 INFO test.py line 173 9944] Loaded configuration file config/ade20k-hrnetv2.yaml [2019-10-30 15:33:25,356 INFO test.py line 174 9944] Running with...

sarimmehdi

Hello, everyone. I tried to benchmark the fps on Titan Xp. using the following command line: > python3 eval.py --cfg config/ade20k-mobilenetv2dilated-c1_deepsup.yaml But it turns out the fps was very low...

UsedToBe97

Class number and name association

1

Hey I am unable to understand how to class values in mask correspond to class names. I want to only use a subset of classes and for that need to...

bluesky314

Replace lines 56-56 in "test.py" with the following: ``` # scores = torch.zeros(1, cfg.DATASET.num_class, segSize[0], segSize[1]) # scores = async_copy_to(scores, gpu) ``` to ` scores = torch.zeros(1, cfg.DATASET.num_class, segSize[0], segSize[1],...

hjinlee88

Does it have cascade segmentation module?

1

Hello, in the paper you said you propose a cascade segmentation module, and this module can be potentially be integrated in any previous semantic segmentation networks. Did you add the...

lgqfhwy

Inference on multiple gpus

1

Is it possible to perform inference/test on multiple gpus. Right now the test script seems to accepts only one gpu id and testing on large dataset is very slow.

Aradhya0510