Some notes on installation

[taesiri]

Hello,

First of all, thank you for sharing the source code.

I encountered some issues while trying to install Semantic-SAM following the instructions in the README file. In the Semantic-SAM repository, there is a pyproject.toml file, and theoretically, it should take care of the installation. However, due to some strange issues, I had to install it manually.

Here are the steps I followed for anyone who is interested:

Using a fresh Anaconda environment to install Semantic-SAM:

pip3 install torch==1.13.1 torchvision==0.14.1 --extra-index-url https://download.pytorch.org/whl/cu113
python -m pip install 'git+https://github.com/MaureenZOU/detectron2-xyz.git'
pip install git+https://github.com/cocodataset/panopticapi.git
git clone https://github.com/UX-Decoder/Semantic-SAM
cd Semantic-SAM
python -m pip install -r requirements.txt

After this, you need to compile and install MultiScaleDeformableAttention:

git clone https://github.com/facebookresearch/Mask2Former
cd Mask2Former/mask2former/modeling/pixel_decoder/ops
sh make.sh

Additionally, you need to install mpi4py:

conda install -c conda-forge mpi4py

You can install SEEM and SAM using the pip+git method:

# Install SEEM
pip install git+https://github.com/UX-Decoder/Segment-Everything-Everywhere-All-At-Once.git@package
# Install SAM
pip install git+https://github.com/facebookresearch/segment-anything.git

I hope this helps resolve any installation issues.

[jwyang]

Hi, @taesiri, really appreciate for your instruction to install the package! This is very useful, I will go to handle the inconvenience of installation.

Best.
Jianwei

[jwyang]

Already fixed, closing this issue.

[dddraxxx]

Hello,

First of all, thank you for sharing the source code.

I encountered some issues while trying to install Semantic-SAM following the instructions in the README file. In the Semantic-SAM repository, there is a pyproject.toml file, and theoretically, it should take care of the installation. However, due to some strange issues, I had to install it manually.

Here are the steps I followed for anyone who is interested:

Using a fresh Anaconda environment to install Semantic-SAM:

pip3 install torch==1.13.1 torchvision==0.14.1 --extra-index-url https://download.pytorch.org/whl/cu113
python -m pip install 'git+https://github.com/MaureenZOU/detectron2-xyz.git'
pip install git+https://github.com/cocodataset/panopticapi.git
git clone https://github.com/UX-Decoder/Semantic-SAM
cd Semantic-SAM
python -m pip install -r requirements.txt

After this, you need to compile and install MultiScaleDeformableAttention:

git clone https://github.com/facebookresearch/Mask2Former
cd Mask2Former/mask2former/modeling/pixel_decoder/ops
sh make.sh

Additionally, you need to install mpi4py:

conda install -c conda-forge mpi4py

You can install SEEM and SAM using the pip+git method:

# Install SEEM
pip install git+https://github.com/UX-Decoder/Segment-Everything-Everywhere-All-At-Once.git@package
# Install SAM
pip install git+https://github.com/facebookresearch/segment-anything.git

I hope this helps resolve any installation issues.

It is weired but I encountered "NameError: name 'MPI' is not defined" for SEEM package and need to add "from mpi4py import MPI" in that file

[taesiri]

@dddraxxx Have you installed mpi4py?

conda install -c conda-forge mpi4py

[dddraxxx]

@dddraxxx Have you installed mpi4py?

conda install -c conda-forge mpi4py

yes, but still need to import that in the 'seem' library code

[taesiri]

@dddraxxx You should check and verify if mpi4py and openmpi are installed.

Btw, I can share my environment.yaml file if you think that can help.

[dddraxxx]

@taesiri yes. I installed both via conda and the code can run well if I add that line from mpi4py import MPI.

[wentaoyuan]

I had the same issue with MPI. It seems that
pip install git+https://github.com/UX-Decoder/Segment-Everything-Everywhere-All-At-Once.git@package
installs the package branch of the SEEM repo which is not fully in sync with v1.0 branch.
Thus, utils/distributed.py is missing an import statement at line 7: https://github.com/UX-Decoder/Segment-Everything-Everywhere-All-At-Once/blob/7b2e76dbb17d0b7831c6813a921fe2bc8de22926/utils/distributed.py#L7.

[NoOneUST]

I had the same issue with MPI. It seems that pip install git+https://github.com/UX-Decoder/Segment-Everything-Everywhere-All-At-Once.git@package installs the package branch of the SEEM repo which is not fully in sync with v1.0 branch. Thus, utils/distributed.py is missing an import statement at line 7: https://github.com/UX-Decoder/Segment-Everything-Everywhere-All-At-Once/blob/7b2e76dbb17d0b7831c6813a921fe2bc8de22926/utils/distributed.py#L7.

I also met the same problem. How did you solve it?

[abrichr]

See #12 (comment) for an explanation and fix.