Visual AI-Based Inventory
Tracking For a Laundromat Operator
Edge AI Based Garment Tracking
Client
Wash and fold laundromat operator
Industry
Laundry Services
Retail Services
Solution
Custom AI Engineering
Edge AI Deployment
Use Case
Garment Tracking and Verification
Loss Prevention
Technology
Visual AI
Computer Vision
Edge Computing
Goal
Develop a custom Visual AI solution for a laundromat operator to track and verify garments throughout the wash and fold workflow, reduce manual errors, and improve operational reliability and customer trust.
Problem Statement
A wash and fold laundromat operator approached Vedya Labs to address garment loss, manual tracking errors, and recurring customer disputes. Processing thousands of garments daily made accurate tracking difficult, while changing garment shapes and compressed surveillance camera footage complicated automated detection. The client needed a solution that improved verification accuracy without disrupting existing operations
Solution Highlights
Vedya Labs designed and deployed a custom Visual AI system tailored to the client’s laundromat operations, integrating IP cameras with NVIDIA based edge computing to verify garments before and after washing.
Garment Detection using Edge AI
Implemented AI models to detect, count, and classify garments during intake and after washing, enabling automated verification at both stages
Multistage Verification
Linked garment records using a common transaction ID to compare before wash and after wash counts and garment types
Key Frame Selection and Classification
Developed custom logic to select the most suitable video frames for garment classification despite continuous changes in garment shape and movement
Hybrid AI Model Approach
Combined YOLO based detection models with cloud-based language model validation to improve garment classification reliability
Optimized Real Time Performance
Optimized the AI pipeline to increase processing speed from approximately 3 FPS to 50 FPS on the NVIDIA Jetson Orin NX platform
Data and Model Engineering
Collected and annotated approximately 90,000 valid frames across multiple laundromat environments to train and evaluate models under varied real world conditions
On Device Processing
Enabled local AI inference on edge hardware to reduce latency and dependence on remote computing infrastructure
Conclusion
Vedya Labs addressed the client’s garment tracking challenges through custom AI engineering, model optimization, and edge deployment. The resulting system improved verification accuracy and processing speed while fitting into existing workflows. At pilot sites, the solution was associated with zero reported garment disputes, demonstrating the value of tailored Visual AI in improving operational transparency and customer confidence.