Contact person: Business Client

Phone:Show

Email:Show

Location: Tallahassee, United States

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"Job Title:
Build AI Garment Detection and Work-Order Printing Prototype for Dry Cleaners (Patent Pending)

Project Summary
I have filed a U.S. Provisional Patent (No. 63/911,322) for an AI system that automatically identifies, counts, and categorizes garments for dry-cleaning operations.

I need an experienced AI and computer-vision developer to build a small-scale working prototype (MVP) that I can demonstrate to SMRT and other POS companies.

Goal: a clerk takes one overhead photo of up to five garments. The AI detects, classifies, and prints a barcoded work order and individual item tags automatically. The system should run smoothly and require no coding by the operator.


Core Objectives

1. Detect and count up to five garments per photo using instance segmentation.
2. Classify garment type: shirt, pants, shorts, dress, jacket, sweater.
3. Identify color: black, white, gray, navy, blue, red, green, brown, tan, other.
4. Determine gender or style: men, women, unisex, unknown.
5. Generate a JSON work order following the provided schema.
6. Print a master invoice and per-item tags in ZPL format with barcodes.


Preferred Technical Stack
Detection: YOLOv8 or YOLO11-seg or equivalent
Automation: Make (Integromat) or Zapier connected to Google Drive, Google Sheets, and PrintNode or QZ Tray
Demo App: Streamlit or Gradio with optional offline ONNX runtime
Printing: Zebra ZD-series or compatible thermal printer
Mock POS API: endpoints /workorders and /print with Swagger and Postman documentation


**Deliverables**

1. Trained detection and classification models (mAP ≥ 0.85, color accuracy ≥ 0.90)
2. Working end-to-end automation from photo to printed labels in under 6 seconds
3. Local demo app that allows detect, edit, and print workflow offline
4. Mock POS REST API with tested documentation
5. Short demonstration video showing toss, detect, and print sequence
6. Signed NDA and IP assignment referencing the patent number below



Legal and IP
All work is work-for-hire and owned by the inventor.
This invention is patent pending:
U.S. Provisional Patent Application No. 63/911,322, filed November 4, 2025.
Contractor must sign an NDA and IP-transfer agreement before starting work.


Budget and Timeline
Fixed Price: 750-1.5kUSD
Delivery Time: 10 to 14 days
Bonus available for early or above-target completion.


Preferred Skills
Experience with computer vision and object detection (YOLO or equivalent)
Python development using Streamlit or Gradio
No-code automation (Zapier or Make)
Label printing in ZPL and PrintNode integration
POS or inventory system integration experience


How to Apply
Provide the following in your proposal:

1. Example of previous object-detection or segmentation work.
2. Plan for handling overlapping garments and color accuracy.
3. Short implementation outline and estimated timeline.
4. Confirmation that you will sign an NDA referencing U.S. Provisional Patent Application No. 63/911,322.


Patent Pending – U.S. Provisional Patent Application No. 63/911,322 (filed November 4, 2025)" (client-provided description)