Collect 1 Million Home Depot and Lowe’s Google Maps Photos

Upwork prohibits unpaid custom test work, so the capability check below is a paid, funded milestone within the $500 total budget. The posting also tells applicants not to perform the test before a contract and funded milestone are in place. ⸻ Collect 1 Million Home Depot and Lowe’s Google Maps Photos Budget $500 total fixed price, all inclusive This project is intended for a contractor who already has scalable image-collection technology and can complete the work within the stated budget. Project Objective Collect and organize 1,000,000 unique photos associated with Home Depot and Lowe’s U.S. store locations: * 700,000 Home Depot photos * 300,000 Lowe’s photos * 1,000,000 total photos Target date range: August 1, 2021 through August 1, 2026 Attempt every Home Depot and Lowe’s U.S. location. The solution must be reusable for additional retailers, future images, and recurring updates. Paid Proof of Capability Do not perform or submit custom test work with your proposal. The selected contractor will first receive a $10 funded milestone to demonstrate the collection process using these two locations: Home Depot Cumberland #0121 2450 Cumberland Parkway Atlanta, GA 30339 Lowe’s Mooresville #0595 509 River Highway Mooresville, NC 28117 For the proof of capability, collect all available photos associated with both store locations and deliver: * All collected image files * All available image metadata * Store and location information * Source photo ID and URL * Uploader name and public profile information * Upload or publication date when available * Caption, review, rating, and associated text when available * Image width, height, file size, and collection date * Exact-duplicate and near-duplicate identification * CSV or JSON metadata export * Summary counts for each store * Sample OCR and image classification results * Demonstration that the process is automated and restartable The proof of capability is only an evaluation. It does not reduce the one-million-image production requirement unless approved by the client. After the capability milestone is accepted, the remaining $490 production milestone will be activated. Production Requirements Collect and deliver: * 700,000 unique Home Depot images * 300,000 unique Lowe’s images * All available associated metadata * Coverage across all U.S. store locations * Images from the requested five-year period * Raw images and useful thumbnails * Exact-duplicate removal * Near-duplicate grouping * Structured metadata exports * Source code and operating instructions An image counts toward the production target when it: * Opens correctly and is not corrupt * Is associated with the correct retailer and store * Falls within the requested period based on available information * Has a valid checksum * Is not an exact duplicate * Includes all available source metadata * Is delivered to the agreed Azure storage location Metadata to Capture Capture all available fields, including: * Retailer * Store number * Store name * Store address * Latitude and longitude * Google Maps listing or Place ID * Source photo ID * Source photo URL * Source review or post URL * Uploader display name * Uploader public profile URL * Uploader public avatar URL * Upload or publication date * Original relative-date text * Caption or description * Associated review text * Review rating * Helpful or reaction count * Image width and height * File type and file size * Collection timestamp * SHA-256 checksum * Perceptual hash * Exact-duplicate group * Near-duplicate group * Processing status and errors Unavailable fields should be left blank rather than estimated without a clear basis. OCR and Image Recognition The processing workflow should support: * Interior-versus-exterior classification * Product aisle and product bay identification * Department classification * Brand and logo recognition * Shelf-label OCR * Price-label OCR * Promotional-sign OCR * Product-name recognition * Model number, SKU, and UPC extraction * Product-package detection Use whole-image OCR first. For likely aisle, shelf, bay, or product images, use overlapping high-resolution crops or tiles to recognize smaller text. Save: * Recognized text * Confidence * Bounding coordinates * OCR engine and model * Detected brands * Detected department * Product-match candidates when possible Azure Delivery Deliver the completed dataset to a client-controlled Azure Blob Storage account. Recommended folders: * raw-images * thumbnails * metadata * ocr * classifications * duplicate-groups * collection-runs * errors Also provide metadata in CSV, JSON, Parquet, or a documented database format. Costs The contractor must include all project costs in the $500 fixed price, including: * APIs and data sources * Compute * OCR * Image recognition * Temporary storage * Azure upload and storage during the project * Bandwidth and data transfer * Quality checks * Development and testing * Documentation and handoff The client will not separately reimburse API, cloud, compute, OCR, bandwidth, or other third-party expenses. The contractor is responsible for compliance with all applicable requirements and source-platform terms. Additional Images and Updates In the proposal, also quote future fixed prices for: * Each additional 100,000 Home Depot images * Each additional 100,000 Lowe’s images * Each additional mixed batch of 100,000 images * Monthly refresh of all stores * Quarterly refresh of all stores * Annual refresh of all stores * Reprocessing existing images with improved OCR * Adding another retailer These future services are not included in the current $500 project unless specifically stated. Deliverables * Two-store paid proof-of-capability dataset * One million production images * 700,000 Home Depot images * 300,000 Lowe’s images * Store master list * Complete available metadata * Raw images and thumbnails * Duplicate groups * OCR and classification outputs * CSV, JSON, Parquet, or database exports * Azure delivery * Complete source code * Installation and operating instructions * Incremental-update capability * Error and coverage reports Proposal Instructions Begin your proposal with: ONE MILLION RETAIL PHOTOS Then answer: 1. Confirm that you accept the $500 total fixed-price budget. 2. Confirm that no API, compute, OCR, storage, or other expenses will be billed separately. 3. Confirm that you will not perform custom test work until the paid capability milestone is funded. 4. Explain how you will collect all photos from the two specified test stores. 5. Estimate the number of available photos for each test store. 6. Describe the metadata you can capture. 7. Describe how the process will scale to one million images. 8. Describe exact-duplicate and near-duplicate detection. 9. Describe the OCR and image-recognition process. 10. State the expected storage required for one million images. 11. Describe the Azure delivery structure. 12. Provide pricing for additional 100,000-image batches. 13. Provide monthly, quarterly, and annual refresh pricing. 14. Provide an example of similar high-volume image-collection work. 15. Confirm that all code, data, metadata, and documentation will be transferred to the client. 16. Confirm that you accept responsibility for compliance with all applicable requirements. Generic proposals that do not address the two proof-of-capability locations will not be considered. Most recent photos first.