Our Solution
Waste Plastic Recycling
We build a complete resource circulation ecosystem for waste plastics centered on PET bottles.
By combining AI-based real-time sorting technology and on-site preprocessing systems (shredding), we fundamentally prevent mixed disposal,
and establish a resource circulation flow that enables the production of high-quality rPET. Through integrated solutions covering both private and public sectors, we produce clean recycled plastic (rPET), simultaneously achieving environmental protection and resource efficiency.
Core Business Areas
DeegoLabs rPET Resource Circulation Solution
Integrated Private + Public
Resource Circulation Infrastructure
AI Sorting-Based On-Site Resource Recovery
- Edge computing-based AI vision technology detects and classifies PET bottles in real-time at over 50fps per second.
- Ensures sorting performance with over 99% accuracy even in challenging environments considering position, rotation, and overlap.
- Solves the problem of household waste mixing to reduce recycling process costs and contamination rates.
Preprocessing (Shredding)-Based High-Quality Recycling
- Sorted PET is immediately shredded into 30×10mm flakes on-site to improve transportation efficiency and quality.
- Filters for contaminant removal and label residue through AI-based dual verification (precision over 0.98) to secure only clean PET.
Integrated Private and Public Resource Circulation System
DeegoLabs provides two solutions optimized for private and public environments respectively.
Private Sector: AI Sorter + Shredder Integrated System
AI sorters installed at corporate sites automatically classify PET, sorting and compressing only recyclable PET for systematic collection.
Public Sector: Reverse Vending Machine (RVM) Integration
When citizens directly deposit PET bottles, AI immediately determines their condition, sorting and compressing only recyclable PET for systematic collection.
Building a mixing-prevention resource circulation infrastructure by combining both systems
Ultra-High-Speed, High-Precision AI Vision Sorting Technology
Fast Processing Speed
Achieves the following frame rates in various industrial environments
| Application Area | Average Processing Speed | Notes |
|---|---|---|
| Industrial AI Vision (Classification/Inspection) | 10~30fps | Mainly label inspection, defect detection, etc. |
| Real-time Object Detection (YOLOv5~8, SSD, etc.) | 20~60fps | High speed possible with GPU |
| High-Speed Line Inspection (FA/Robot Vision) | 40~120fps | Based on line scan camera |
| Commercial-Level Waste Plastic Sorter | 20~40fps | Average around 30fps considering bottle position, rotation, overlap |
- Real-time Detection: Over 50fps
- Edge computer processing, not expensive server PCs
- Even at high speeds, guarantees object tracking success rate (recall 0.99)
High Precision
Applies the following advanced algorithms for accurate PET separation and verification.
Contaminant-Label Classification Algorithm
- Utilizes ECA (Efficient Channel Attention) technique, updating weights according to each channel's importance
- Effect of making the model focus on important features
Filtering Technique for Verifying Clean PET
- VAE (Variational Auto Encoder) recon error threshold filtering
- Through dual verification to ensure impurity-free PET, achieves precision of 0.98 or higher
