Update additional requirements for model init --streaming-video

[mimarcel]

Why

• new package shapely is required for polygon filtering in streaming videos

How

• Instead of hardcoding the new package in the list of requirements, we suggest to install clarifai-protocol with the auto-annotation extra that will dynamically include the necessary requirements for video livestreaming

Tests

clarifai model init --streaming-video

Initializing model with default template...

  Model initialized in /Users/abc/my-model

  1. Edit 1/model.py with your model logic
  2. Add dependencies to requirements.txt
  3. Customize Dockerfile if needed (auto-regenerated on upload if deleted)

  Test locally:
    clarifai model serve /Users/abc/my-model
    clarifai model serve /Users/abc/my-model --mode env       # auto-create venv and install deps
    clarifai model serve /Users/abc/my-model --mode container # run inside Docker

  Deploy to Clarifai:
    clarifai model deploy /Users/abc/my-model --instance gpu-nvidia-a10g
    clarifai list-instances                             # list available instances

Notes

• the auto-annotation extra was introduced in clarifai-protocol version 0.0.60, as documented in https://pypi.org/project/clarifai-protocol/0.0.60.

[github-actions]

Package	Line Rate	Health
clarifai	45%	❌
clarifai.cli	61%	➖
clarifai.cli.templates	67%	➖
clarifai.cli.templates.toolkits	100%	✔
clarifai.client	65%	➖
clarifai.client.auth	67%	➖
clarifai.constants	100%	✔
clarifai.datasets	100%	✔
clarifai.datasets.export	69%	➖
clarifai.datasets.upload	75%	➖
clarifai.datasets.upload.loaders	37%	❌
clarifai.models	100%	✔
clarifai.rag	0%	❌
clarifai.runners	52%	➖
clarifai.runners.models	58%	➖
clarifai.runners.pipeline_steps	39%	❌
clarifai.runners.pipelines	72%	➖
clarifai.runners.utils	62%	➖
clarifai.runners.utils.data_types	72%	➖
clarifai.schema	100%	✔
clarifai.urls	58%	➖
clarifai.utils	65%	➖
clarifai.utils.evaluation	16%	❌
clarifai.workflows	95%	✔
Summary	60% (11357 / 18957)	➖

Minimum allowed line rate is 50%

[josh-tepper]

NOTE, the Dockerfile template; after the ${ADDITIONAL_PACKAGES} insertion -- which includes this test -- we subsequently install the requirements file:

clarifai-python/clarifai/runners/dockerfile_template/Dockerfile.template

Line 47 in e9510d1

RUN ["uv", "pip", "install", "--no-cache-dir", "-r", "/home/nonroot/requirements.txt"]

Concern: if a user defines a different version of clarify protocol in their local requirements.txt, the subsequent installation of their version could override the initial install, causing the add-ons to be lost.

Could you please test:

• whether those two extra packages actually remain after the installation from requirements.txt?
• If there any way to install only the add-ons and not the main package itself?

[mimarcel]

Very good point! I tested this and it is not a problem,: the additional packages remain installed after installing a different version of the clarifai-protocol package.

Test details follow.

1. install clarifai-protocol with auto-annotation extra:

   uv pip install 'clarifai-protocol[auto-annotation]==0.0.60'

2. check additional packages installed

   uv pip list | grep -E "shapely|av"
   av                                       16.1.0
   shapely                                  2.1.2

3. install different clarifai-protocol version without auto-annotation extra:

   uv pip install 'clarifai-protocol==0.0.59'

4. check additional packages installed

   uv pip list | grep -E "shapely|av"
   av                                       16.1.0
   shapely                                  2.1.2