numpy 2.0

[johnnynunez]

MediaPipe Solution (you are using)

main

Programming language

python

Are you willing to contribute it

None

Describe the feature and the current behaviour/state

Actual frameworks support numpy 2.0... here is missing support

Will this change the current API? How?

No response

Who will benefit with this feature?

No response

Please specify the use cases for this feature

mediapipe breaks due that is not supporting 2.0 in a lot of plugins and frameworks

Any Other info

No response

[HarishPermal]

Hi @johnnynunez,

Thanks for reporting this. To help us understand and reproduce the issue, could you please provide a few more details?

• Which plugins or frameworks are you using that are causing the break?

• Could you please share a minimal code sample and the full error message you're seeing?

This will help us pinpoint the problem.

Thanks 😊

[johnnynunez]

Hi @johnnynunez,

Thanks for reporting this. To help us understand and reproduce the issue, could you please provide a few more details?

• Which plugins or frameworks are you using that are causing the break?
• Could you please share a minimal code sample and the full error message you're seeing?

This will help us pinpoint the problem.

Thanks 😊

Code: https://github.com/johnnynunez/jetson-containers/tree/master/packages/diffusion/stable-diffusion-webui

#---
# name: stable-diffusion-webui
# group: diffusion
# depends: [pytorch, torchvision, transformers, xformers, pycuda, opencv, onnxruntime]
# requires: '>=34.1.0'
# docs: docs.md
# notes: https://github.com/AUTOMATIC1111/stable-diffusion-webui
#---
ARG BASE_IMAGE
FROM ${BASE_IMAGE}

ARG STABLE_DIFFUSION_WEBUI_REF \
    STABLE_DIFFUSION_WEBUI_SHA \
    AUTOMATIC1111_ROOT_DIR="/opt/stable-diffusion-webui"

ADD https://api.github.com/repos/AUTOMATIC1111/stable-diffusion-webui/git/${STABLE_DIFFUSION_WEBUI_REF} /tmp/stable_diffusion_webui_version.json

COPY install_extensions.sh /tmp/

RUN set -ex \
    && git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui /opt/stable-diffusion-webui \
    && git -C /opt/stable-diffusion-webui checkout ${STABLE_DIFFUSION_WEBUI_SHA} \
    \
    # Fix: ControlNet init warning: Unable to install insightface automatically. Please try run `pip install insightface` manually. \
    # Fix: https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/13236 \
    && PYTHONPATH="$AUTOMATIC1111_ROOT_DIR" pip3 install -U httpx==0.24.1 GitPython \
    && PYTHONPATH="$AUTOMATIC1111_ROOT_DIR" /tmp/install_extensions.sh \
        https://github.com/dusty-nv/stable-diffusion-webui-tensorrt \
        https://github.com/Mikubill/sd-webui-controlnet \
        https://github.com/zanllp/sd-webui-infinite-image-browsing \
        https://github.com/ahgsql/StyleSelectorXL \
        https://github.com/thomasasfk/sd-webui-aspect-ratio-helper \
        https://github.com/BlafKing/sd-civitai-browser-plus \
    && PYTHONPATH="$AUTOMATIC1111_ROOT_DIR" python3 -c 'from modules import launch_utils; launch_utils.args.skip_python_version_check=True; launch_utils.prepare_environment()' \
    # partially initialized module 'cv2' has no attribute 'gapi_wip_gst_GStreamerPipeline' \
    && /tmp/opencv/install.sh \
    # TypeError: 'type' object is not subscriptable \
    && sed 's|dict\[str, OptionInfo\]|dict|g' -i $AUTOMATIC1111_ROOT_DIR/modules/options.py \
    && sed 's|dict\[.*\]|dict|g' -i $AUTOMATIC1111_ROOT_DIR/modules/processing.py \
    && sed 's|list\[.*\]|list|g' -i $AUTOMATIC1111_ROOT_DIR/modules/processing.py \
    && sed 's|dict\[.*\]|dict|g' -i $AUTOMATIC1111_ROOT_DIR/modules/script_callbacks.py \
    && sed 's|list\[.*\]|list|g' -i $AUTOMATIC1111_ROOT_DIR/modules/script_callbacks.py \
    \
    # test that it loads \
    && cd /opt/stable-diffusion-webui \
    && pip3 install --force-reinstall "git+https://github.com/deepinsight/insightface.git#egg=insightface&subdirectory=python-package" \
    && pip3 install diffusers==0.30 \
    && PYTHONPATH="$AUTOMATIC1111_ROOT_DIR" python3 launch.py --help \
    && PYTHONPATH="$AUTOMATIC1111_ROOT_DIR" pip3 show insightface \
    && PYTHONPATH="$AUTOMATIC1111_ROOT_DIR" python3 -c 'import insightface; print(insightface.__version__);' \
    \
    # Cleanup \
    && rm -rf /var/lib/apt/lists/* \
    && apt-get clean

# set the cache dir for models
ENV DIFFUSERS_CACHE=/data/models/diffusers

# Add the correct link for stable diffusion download
#RUN curl -L https://huggingface.co/Comfy-Org/stable-diffusion-v1-5-archive -o /data/models/stable-diffusion

# default start-up command
CMD ["/bin/bash", "-c", "cd /opt/stable-diffusion-webui && python3 launch.py --data=/data/models/stable-diffusion --enable-insecure-extension-access --xformers --listen --port=7860"]

[Borogravian]

You may want to link with #5612

This constraint on numpy<2 also causes issues with ComfyUI plugins that uses mediapipe, it downgrades numpy for the main app. ComfyUI at https://github.com/comfyanonymous/ComfyUI .

[gitlsl]

Numpy<2 is really old. It is common to use OpenCV and MediatPipe in the same project. When I installed the latest OpenCV, its numpy version was already 2.2.6