Contact person: Business Client

Phone:Show

Email:Show

Location: ALEXANDRIA, United States

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I’m building a unified web and mobile application that lets end-users upload or import their own photos and have every visible item automatically detected, tagged, classified, and grouped in seconds. Accuracy is critical—models should reach near-perfect performance, delivering a clear percentage-based confidence score for each object as well as an overall match consistency when similar items appear across multiple images.

The workflow I have in mind: images arrive through a responsive browser and companion mobile interface, flow through an AI/ML pipeline (TensorFlow, PyTorch, YOLO-v8 or an equivalent state-of-the-art model), and return annotated results to the screen. A downloadable report (PDF and CSV) must be generated at the same time, summarizing every object, its class, and its confidence level.

Key deliverables
• Cross-platform front end (React/Vue or Flutter/React Native) and a scalable backend API
• Custom-trained detection model that sustains ≥95 % mAP on my validation set
• Similarity engine producing a percentage likelihood score between objects across images
• Real-time annotation overlay plus a clean PDF/CSV report template
• Cloud deployment (AWS, GCP or Azure) with CI/CD, containerisation, and full source code & documentation

Acceptance criteria
• End-to-end inference under one second for a 2 K image on both web and mobile
• Model evaluation metrics and confusion matrix included in the technical report
• Successful test run on a hold-out image set showing ≥95 % precision and recall

Once these milestones are met, I’ll sign off and move to the next phase—continuous learning and incremental feature releases." (client-provided description)