Install supervision with numpy<2

[pirnerjonas]

Search before asking

• I have searched the Supervision issues and found no similar feature requests.

Question

Hi,

I wanted to add supervision to a project of mine that depends on numpy<2.

If I try this via poetry I get:

(.venv) C:\Users\z0039hdz\Projects\chip-segmentation [feat/export-model ≡ +3 ~3 -0 !]> poetry env info

Virtualenv
Python:         3.10.11
Implementation: CPython
Path:           C:\Users\z0039hdz\Projects\chip-segmentation\.venv
Executable:     C:\Users\z0039hdz\Projects\chip-segmentation\.venv\Scripts\python.exe
Valid:          True

Base
Platform:   win32
OS:         nt
Python:     3.10.11
Path:       C:\Users\z0039hdz\.pyenv\pyenv-win\versions\3.10.11
Executable: C:\Users\z0039hdz\.pyenv\pyenv-win\versions\3.10.11\python.exe

(.venv) C:\Users\z0039hdz\Projects\chip-segmentation [feat/export-model ≡ +3 ~3 -0 !]> poetry add supervision
Using version ^0.25.1 for supervision

Updating dependencies
Resolving dependencies... (5.3s)

Because no versions of supervision match >0.25.1,<0.26.0
 and supervision (0.25.1) depends on numpy (>=2.1.0), supervision (>=0.25.1,<0.26.0) requires numpy (>=2.1.0).
So, because chip-segmentation depends on both numpy (<2) and supervision (^0.25.1), version solving failed.

it seems like supervision depends on numpy>=2.1.0 for python 3.10 but if I check the pyproject.toml of supervision I see "numpy>=1.21.2".

Is it possible to use supervision with numpy<2? Am I doing something wrong?

Additional

No response

[onuralpszr]

@pirnerjonas hello in current "released" version of supervision If I create a new python project with toml and add numpy<2 and latest supervision like this ;

[project]
name = "test-env"
version = "0.1.0"
description = "Add your description here"
readme = "README.md"
requires-python = ">=3.10"
dependencies = [
    "numpy<2",
    "supervision>=0.25.1",
]

I was able to install without problem. You might want to check your other packages or conflicts.

[onuralpszr]

Converting to Q&A Discussion

[Shahram-73]

This looks like a dependency resolver conflict rather than supervision requiring NumPy 2.x in general. In a clean env, numpy<2 + supervision can resolve, but another dependency in your lock graph may be forcing numpy>=2.x.
GitHub

What I’d try:
Create a totally clean env and install just these two:

python -m venv .venv
. .venv/bin/activate # (or .venv\Scripts\activate on Windows)
pip install "numpy<2" "supervision==0.25.1"

If that works, add your other deps back one by one to find who pulls NumPy 2.x.
For Poetry specifically, poetry show --tree | findstr numpy (Windows) / grep numpy (mac/linux) helps identify the package that introduces the constraint.
If you paste your poetry.lock snippet around NumPy (or poetry show --tree), it’s usually obvious which package is forcing the bump.