Feature/make the sowlv2 more efficient

[bladeszasza]

No description provided.

[bladeszasza]

This pull request introduces significant enhancements to the SOWLv2 pipeline, focusing on temporal video tracking, V-JEPA 2 integration, optimized inference, and documentation updates. The changes aim to improve the pipeline's efficiency, scalability, and usability, particularly for video processing tasks. Below are the key updates grouped by theme.

Temporal Video Tracking and V-JEPA 2 Integration

• Added a comprehensive temporal video tracking system leveraging V-JEPA 2 for intelligent frame selection, OWLv2 for multi-frame detection, and SAM2 for consistent object tracking. This system enables efficient and accurate object detection across key frames while reducing computational overhead. (docs/temporal_video_tracking.md, docs/temporal_video_tracking.mdR1-R253)
• Integrated V-JEPA 2 into the SOWLv2 pipeline to enhance temporal understanding, motion handling, and frame selection. Includes new methods for feature extraction, frame importance scoring, and optimized frame selection. (docs/vjepa2_integration.md, docs/vjepa2_integration.mdR1-R256)

Optimized Inference

• Introduced an example script for optimized inference using parallel processing and GPU optimization. The script demonstrates benchmarking, comparison with the standard pipeline, and GPU memory profiling. (examples/optimized_inference.py, examples/optimized_inference.pyR1-R176)

Documentation Improvements

• Enhanced the README.md with a structured table for command-line options and improved descriptions for configuration file usage. (README.md, [1] [2]

Configuration and Testing Updates

• Updated .claude/settings.local.json to include additional test cases and commands for integration testing, such as new pytest commands for optimization tests and HuggingFace domain fetches. (.claude/settings.local.json, .claude/settings.local.jsonL45-R52)

Version Increment

• Bumped the project version from 0.2.1 to 0.2.2 in pyproject.toml to reflect the new features and improvements. (pyproject.toml, pyproject.tomlL7-R7)

[Copilot]

Pull Request Overview

This PR introduces a new optimized pipeline for SOWLv2 that replaces the legacy SOWLv2Pipeline with the more efficient OptimizedSOWLv2Pipeline along with extensive parallel and GPU optimizations.

• Replaces direct instantiation of SOWLv2Pipeline with OptimizedSOWLv2Pipeline using ParallelConfig in tests, CLI, and internal modules.
• Adds new integration tests and optimization modules (e.g. V-JEPA 2 integration, GPU and batch optimizations) to improve processing efficiency.
• Updates documentation, CLI arguments, and version numbers to support the new optimization features.

Reviewed Changes

Copilot reviewed 20 out of 25 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
tests/integration/*	Updated tests to use OptimizedSOWLv2Pipeline with ParallelConfig and new mocks for multi-prompt detection.
sowlv2/optimizations/*	Added modules to support parallel processing, GPU optimizations, intelligent batching, and V-JEPA 2 integration.
sowlv2/cli.py	Modified CLI to default to the optimized pipeline and added new optimization-related arguments.
setup.py, pyproject.toml	Updated version to reflect new optimized release.
docs/*	Added documentation for V-JEPA 2 integration and temporal video tracking.

[Copilot]

The repeated instantiation of ParallelConfig in multiple tests could be refactored into a shared fixture to reduce duplication and improve maintainability.

Suggested change

	pipeline = OptimizedSOWLv2Pipeline(pipeline_config, ParallelConfig())
	pipeline = OptimizedSOWLv2Pipeline(pipeline_config, parallel_config)

[Copilot]

Consider replacing print statements with a logging framework to enable configurable output and better integration with production environments.

Suggested change

	print("Using optimized SOWLv2 pipeline...")
	logging.info("Using optimized SOWLv2 pipeline...")

[github-advanced-security]

This pull request sets up GitHub code scanning for this repository. Once the scans have completed and the checks have passed, the analysis results for this pull request branch will appear on this overview. Once you merge this pull request, the 'Security' tab will show more code scanning analysis results (for example, for the default branch). Depending on your configuration and choice of analysis tool, future pull requests will be annotated with code scanning analysis results. For more information about GitHub code scanning, check out the documentation.