|
Smartphone-based OCR and computer vision system for outbound product checking in warehouses |
|---|---|
| รหัสดีโอไอ | |
| Title | Smartphone-based OCR and computer vision system for outbound product checking in warehouses |
| Creator | Sirawich Ngernsalung |
| Contributor | Warut Pannakkong, Advisor |
| Publisher | Thammasat University |
| Publication Year | 2568 |
| Keyword | Computer vision, Optical character recognition, Object detection, Deep learning, Warehouse management |
| Abstract | Outbound product verification in warehouses is often performed using handheld barcode scanners, which can be slow and physically demanding, especially for multilabel items. This thesis presents a smartphone-based system that uses object detection (YOLOv8) and OCR (PaddleOCR) to streamline the verification process. Developed entirely with open-source tools, the system is cost-effective and suitable for small and medium-sized enterprises (SMEs). Tested on 2,331 real-world warehouse images, the system achieved 100% label detection and 98.07% OCR accuracy. A mock-up experiment showed it reduced task completion time by 29% to 91% compared to barcode scanning, with all improvements statistically significant. Unlike barcode systems, its performance remains consistent regardless of label count. The system also reduces operator fatigue, requires no specialized hardware, and supports flexible deployment. These findings suggest it is a reliable, efficient, and scalable alternative to conventional outbound checking methods. |
