[Feature]: Add Docker support for reproducible setup and easier onboarding.

[priyan17singh]

Summary

This proposal suggests adding official Docker support to provide a consistent, reproducible environment for running PerceptionMetrics (CLI and GUI), reducing setup complexity and platform-specific issues.

---

❗ Problem

The project relies on several heavy and system-dependent libraries:

• PyTorch
• OpenCV
• Open3D
• pycocotools
• ultralytics

Setting up these dependencies manually can lead to:

• Platform-specific errors (e.g., DLL issues on Windows).
• Missing system libraries in Linux environments.
• Inconsistent behavior across machines.
• Increased onboarding time for new contributors.

For example, when attempting to run the project in a minimal Linux/Windows environment, the application failed due to missing OpenGL libraries required by Open3D:

OSError: libGL.so.1: cannot open shared object file.
OSError: [WinError 1114] A dynamic link library (DLL) initialization routine failed.

This indicates that system-level dependencies are not explicitly managed.

---

Proposed Solution

Introduce containerized execution using Docker to standardize the runtime environment.

Scope of changes

• Add a Dockerfile for CPU-based execution.
• (Optional) Add Dockerfile.gpu for CUDA-enabled systems.
• Add .dockerignore to optimize build context.
• Update README with Docker usage instructions.

---

Impact

This enhancement improves developer experience, usability, and reliability of the project without affecting existing workflows.

---