Nilaksh Parihar@pariharnilaksh26000

AMD Captivate AI is an end-to-end intelligent video captioning and understanding system built for the AMD Developer Hackathon 2026. The project automates the creation of high-quality descriptive captions tailored to different target audiences. On container startup, the batch processor dynamically reads video URLs from an input file, downloads them, and utilizes computer vision (OpenCV) to perform intelligent keyframe extraction. By applying a grayscale structural difference algorithm, it automatically filters out near-duplicate frames to minimize token usage and focus only on unique, representative visual changes. These keyframes are captioned using the Fireworks AI Qwen 3.5/3.7 Plus Vision model. The individual frame descriptions are consolidated to generate a single, cohesive overall video summary. Finally, the system rewrites this summary into four distinct styles to fit different contexts: Formal (professional and objective), Sarcastic (witty and mocking), Humorous-Tech (loaded with coding and tech references), and Humorous-NonTech (everyday humor). The results are written into a structured JSON file matching the Track 2 specifications. The system also supports a responsive React dashboard for local deployment, featuring SRT subtitle exporting, visual keyframe timelines, and styled captions.

To participate in the hackathon, simply click Sign up with AMD. If you are not already a member of the AMD AI Developer Program (ADP), you will need to create an account. ADP members can access AI tutorials, experts, and community support. In addition, new members can claim a free month of DeepLearning.ai pro, $100 in AMD GPU cloud credits, and $50 in Fireworks AI API credits. All hackathon participants will receive on hackathon launch additional compute and API credits for their projects. Ready to build what's next on AMD? Sign up with AMD below to secure your spot.