fix: update pcb-aoi dependencies for modern Python compatibility

[aryunewaskar77-art]

What this PR does

Updates outdated dependencies in examples/pcb-aoi/requirements.txt that cause installation failures on modern Python (3.9+).

Problem

Two dependencies were broken:

1. opencv-python~=3.4

• Version 3.4 has no pre-built wheels for Python 3.9+
• Forces a source build which fails on modern macOS and most Linux systems
• Blocks any new contributor from running this example

2. onnx==1.9.0

• Released in 2021, incompatible with numpy>=1.20+
• Causes import errors on modern environments

Changes

• opencv-python~=3.4 → opencv-python>=4.5.0,<4.10.0
• onnx==1.9.0 → onnx>=1.12.0
• Added explicit numpy>=1.21.0 dependency

Verification

Tested on:

• macOS (Apple Silicon)
• Python 3.14.2

All three packages install successfully with no errors.

Related issues

Related to #230

[kubeedge-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: aryunewaskar77-art
To complete the pull request process, please assign moorezheng after the PR has been reviewed.
You can assign the PR to them by writing /assign @moorezheng in a comment when ready.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[gemini-code-assist]

Code Review

This pull request updates the dependency versions in examples/pcb-aoi/requirements.txt to support Python 3.9+, specifically upgrading OpenCV and ONNX and adding NumPy. The review feedback identifies a likely runtime incompatibility between the specified NumPy version and the legacy TensorFlow 1.x code used in the project. It also suggests using opencv-python-headless for better container portability and requests clarification on the OpenCV version upper bound.

[gemini-code-assist]

The requirement numpy>=1.21.0 is likely incompatible with the version of TensorFlow used in this example. The code in basemodel.py (lines 25-26) explicitly uses tensorflow.contrib.slim, which identifies it as TensorFlow 1.x. TensorFlow 1.x is known to break with NumPy 1.20+ due to the removal of several deprecated type aliases (like np.object). While newer NumPy is needed for Python 3.9+ and Apple Silicon support, it will likely cause runtime AttributeError when TensorFlow is used. Please verify if the example has been verified to run (not just install) with this NumPy version.

[gemini-code-assist]

For edge and server-side deployments (typical for KubeEdge), it is recommended to use opencv-python-headless instead of opencv-python. The headless version does not require X11 or other GUI-related system libraries, making it more portable and easier to install in minimal container environments. Since the code only uses imread and imwrite and does not appear to use any GUI features like cv2.imshow, this should be a safe drop-in replacement.\n\nAdditionally, is there a specific reason for the upper bound <4.10.0? If not, consider allowing the latest stable versions (4.10+) or pinning to a specific known-good version for better reproducibility.

opencv-python-headless>=4.5.0,<4.10.0

[aryunewaskar77-art]

/assign @MooreZheng