AI Training Sourcing & Licensing of a Diverse Technical Image Dataset (50,000 Images Across 10 Domains)
Publicada el 2026-07-16
Descripción de la oferta
We are seeking an experienced Computer Vision or Technical Image Sourcing vendor to source, curate, and legally license a large-scale, high-quality, and diverse collection of 50,000 technical images. This dataset will be used for AI model training, evaluation, benchmarking, and dense captioning use cases. The project requires a strict distribution of 5,000 images per domain across the following 10 domains: 1. General Science (Microscopy, lab experiments, molecular diagrams, charts, etc.) 2. Maps & Technical Diagrams (Transit maps, floor plans, network/process flowcharts, etc.) 3. Geospatial / Satellite (Satellite/drone imagery, GIS visualizations, topographic maps, etc.) 4. Agriculture (Crop diseases, plant growth stages, farm machinery, irrigation, etc.) 5. Robotics (Robotic arms, warehouse automation, LiDAR visualizations, etc.) Apart from These ( Urgent) 1. Semiconductors Engineering (Wafer photos, die shots, chip layouts, PCB-mounted chips, etc.) 2. Manufacturing (Production lines, industrial machinery, quality/defect inspections, etc.) 3. Robotic & Automation Engineering 4. Power Systems 5. Telecommunication Strict Quality & Technical Requirements: 1. Image Types: A balanced mix of photographs, technical diagrams, schematics, CAD renderings, illustrations, maps, and relevant screenshots. 2. Resolution: Minimum resolution of 1024 * 1024 pixels or higher. Images must be delivered in their highest available native resolution (no artificial upscaling or excessive compression). 3. Formats: JPEG/JPG for natural photographs; PNG for diagrams, maps, schematics, and text-heavy images. 4. Language: Text within images must be strictly English or have no text at all. 5. Exclusions: NO AI-generated images. NO watermarks. NO duplicates or near-duplicates (e.g., minor angle variations, zoom-only changes, minor cosmetic CAD modifications). 6. Every image must add unique visual value withLicensing & Metadata 7. Expectations: Commercial Rights: All images must be legally cleared and licensed for commercial usage, AI model training, and AI evaluation/benchmarking. 8. Metadata: A corresponding metadata file (CSV/JSON) must be provided containing: Source URL/repository, License type, Commercial rights confirmation, Attribution requirements (if any), and Domain/category tagging. 9. Payment Terms & Proposal Requirements: Pricing Structure: Budget proposals must be submitted on a per-image cost basis. Milestones: Payment milestones will be released only after client review and final approval of delivered batches. To Apply: Please share your domain experience, your general sourcing approach, and a link/sample of similar technical image datasets you have successfully curated in the past. Note : Please Share Sample Images over Google Drive Link BR Shubh Sherawat
Skills
Fuente original: freelancer