About

Overview: We need a computer vision expert to build a tool that detects and tracks objects / logos in videos, using APIs like Azure Video Intelligence, Google Vision, or similar. The goal is to capture when and where logos appear, measure brand visibility across short- and long-form video content. Accuracy, robustness, and practical implementation are critical. We need someone with proven expertise to deliver a high-quality, working solution. Key Deliverables: • Detect objects / logos in video files using the most suitable tools or technologies. • Report timestamps (start / end in ms), bounding box coordinates, and confidence scores. • Export results to clean CSV / Excel. • Provide documentation on setup, usage, and modifying input / output paths. • Generate images of detected objects / logos for manual validation (Optional). Requirements: • Python or other reliable tech solutions are acceptable. • No UI needed - script-based workflows with clear documentation are sufficient. • Scripts should be fully reusable with new reference video inputs and may access APIs from computer vision services (e.g., Azure Video Intelligence, Google Vision, or equivalent). • Handle varied video formats, resolutions, and lengths. • Must work without large set of training image (we may provide small logo samples per video if required). • Robust against different lighting, scenes, and placements. • Vendor to use their own cloud account for dev / testing - final solution must run with our credentials. Notes: • Brand-specific recognition is not required. • We’re open to combining tools for the best results. • Please share examples of similar projects you’ve delivered. • We’ll provide additional information and video samples to candidates.