praeclarumjj3 commited on
Commit
ca96ab1
1 Parent(s): f206f62
Files changed (2) hide show
  1. Dockerfile +1 -4
  2. gradio_app.py +4 -4
Dockerfile CHANGED
@@ -30,18 +30,15 @@ RUN chmod -R 777 $WORKDIR
30
 
31
  COPY requirements.txt $WORKDIR/requirements.txt
32
  RUN pip install --no-cache-dir --upgrade -r $WORKDIR/requirements.txt
 
33
 
34
  COPY . .
35
 
36
  ARG TORCH_CUDA_ARCH_LIST=7.5+PTX
37
 
38
- RUN pip install ninja
39
-
40
  USER root
41
  RUN chown -R user:user $HOME
42
  RUN chmod -R 777 $HOME
43
- RUN chown -R user:user $WORKDIR
44
- RUN chmod -R 777 $WORKDIR
45
 
46
  USER user
47
  RUN ln -s $WORKDIR/oneformer/modeling/pixel_decoder/ops/ $WORKDIR/ && ls && cd ops/ && FORCE_CUDA=1 python setup.py build --build-base=$WORKDIR/ install --user && cd ..
 
30
 
31
  COPY requirements.txt $WORKDIR/requirements.txt
32
  RUN pip install --no-cache-dir --upgrade -r $WORKDIR/requirements.txt
33
+ RUN pip install ninja
34
 
35
  COPY . .
36
 
37
  ARG TORCH_CUDA_ARCH_LIST=7.5+PTX
38
 
 
 
39
  USER root
40
  RUN chown -R user:user $HOME
41
  RUN chmod -R 777 $HOME
 
 
42
 
43
  USER user
44
  RUN ln -s $WORKDIR/oneformer/modeling/pixel_decoder/ops/ $WORKDIR/ && ls && cd ops/ && FORCE_CUDA=1 python setup.py build --build-base=$WORKDIR/ install --user && cd ..
gradio_app.py CHANGED
@@ -187,12 +187,12 @@ def segment(path, task, dataset, backbone):
187
 
188
  title = "OneFormer: One Transformer to Rule Universal Image Segmentation"
189
 
190
- description = "<p font-size: 16px; font-weight: w600; text-align: center'> <a href='https://praeclarumjj3.github.io/oneformer/' target='_blank'>Project Page</a> | <a href='https://arxiv.org/abs/2211.06220' target='_blank'>ArXiv Paper</a> | <a href='https://github.com/SHI-Labs/OneFormer' target='_blank'>Github Repo</a></p>" \
191
- + "<p font-size: 12px; text-align: center' margin: 10px font-weight: w300; text-align: center'> <a href='https://chrisjuniorli.github.io/' target='_blank'>Jiachen Li<sup>*</sup></a> <a href='https://www.linkedin.com/in/mtchiu/' target='_blank'>MangTik Chiu<sup>*</sup></a> <a href='https://alihassanijr.com/' target='_blank'>Ali Hassani</a> <a href='https://www.linkedin.com/in/nukich74/' target='_blank'>Nikita Orlov</a> <a href='https://www.humphreyshi.com/home' target='_blank'>Humphrey Shi</a></p>" \
192
- + "<p text-align: center; font-size: 14px; font-weight: w300;'> \
193
  OneFormer is the first multi-task universal image segmentation framework based on transformers. Our single OneFormer model achieves state-of-the-art performance across all three segmentation tasks with a single task-conditioned joint training process. OneFormer uses a task token to condition the model on the task in focus, making our architecture task-guided for training, and task-dynamic for inference, all with a single model. We believe OneFormer is a significant step towards making image segmentation more universal and accessible.\
194
  </p>" \
195
- + "<p text-align: center; font-size: 14px; font-weight: w300;'> [Note: Inference on CPU may take upto 2 minutes. On a single RTX A6000 GPU, OneFormer is able to inference at more than 15 FPS.</p>"
196
 
197
  # description = "<p style='color: #E0B941; font-size: 16px; font-weight: w600; text-align: center'> <a style='color: #E0B941;' href='https://praeclarumjj3.github.io/oneformer/' target='_blank'>Project Page</a> | <a style='color: #E0B941;' href='https://arxiv.org/abs/2211.06220' target='_blank'>OneFormer: One Transformer to Rule Universal Image Segmentation</a> | <a style='color: #E0B941;' href='https://github.com/SHI-Labs/OneFormer' target='_blank'>Github</a></p>" \
198
  # + "<p style='color:royalblue; margin: 10px; font-size: 16px; font-weight: w400;'> \
 
187
 
188
  title = "OneFormer: One Transformer to Rule Universal Image Segmentation"
189
 
190
+ description = "<p style='font-size: 16px; font-weight: w600; text-align: center'> <a href='https://praeclarumjj3.github.io/oneformer/' target='_blank'>Project Page</a> | <a href='https://arxiv.org/abs/2211.06220' target='_blank'>ArXiv Paper</a> | <a href='https://github.com/SHI-Labs/OneFormer' target='_blank'>Github Repo</a></p>" \
191
+ + "<p style='font-size: 12px; text-align: center' margin: 10px font-weight: w300; text-align: center'> <a href='https://praeclarumjj3.github.io/' target='_blank'>Jitesh Jain<sup>*</sup></a> <a href='https://chrisjuniorli.github.io/' target='_blank'>Jiachen Li<sup>*</sup></a> <a href='https://www.linkedin.com/in/mtchiu/' target='_blank'>MangTik Chiu<sup>*</sup></a> <a href='https://alihassanijr.com/' target='_blank'>Ali Hassani</a> <a href='https://www.linkedin.com/in/nukich74/' target='_blank'>Nikita Orlov</a> <a href='https://www.humphreyshi.com/home' target='_blank'>Humphrey Shi</a></p>" \
192
+ + "<p style='text-align: center; font-size: 14px; font-weight: w300;'> \
193
  OneFormer is the first multi-task universal image segmentation framework based on transformers. Our single OneFormer model achieves state-of-the-art performance across all three segmentation tasks with a single task-conditioned joint training process. OneFormer uses a task token to condition the model on the task in focus, making our architecture task-guided for training, and task-dynamic for inference, all with a single model. We believe OneFormer is a significant step towards making image segmentation more universal and accessible.\
194
  </p>" \
195
+ + "<p style='text-align: center; font-size: 14px; font-weight: w300;'> [Note: Inference on CPU may take upto 2 minutes. On a single RTX A6000 GPU, OneFormer is able to inference at more than 15 FPS.</p>"
196
 
197
  # description = "<p style='color: #E0B941; font-size: 16px; font-weight: w600; text-align: center'> <a style='color: #E0B941;' href='https://praeclarumjj3.github.io/oneformer/' target='_blank'>Project Page</a> | <a style='color: #E0B941;' href='https://arxiv.org/abs/2211.06220' target='_blank'>OneFormer: One Transformer to Rule Universal Image Segmentation</a> | <a style='color: #E0B941;' href='https://github.com/SHI-Labs/OneFormer' target='_blank'>Github</a></p>" \
198
  # + "<p style='color:royalblue; margin: 10px; font-size: 16px; font-weight: w400;'> \