A Two-Stage Detection-Tracking Framework for Stable Apple Quality Inspection in Dense Conveyor-Belt Environments

TL;DR

This paper introduces a two-stage detection and tracking framework for stable, real-time apple quality inspection on conveyor belts, combining object detection, multi-object tracking, and defect classification to improve temporal stability and robustness.

Contribution

The novel integration of detection, tracking, and defect classification with track-level aggregation enhances stability and consistency in automated fruit inspection systems.

Findings

Improved temporal stability over frame-wise methods

Enhanced robustness in dense multi-object environments

Effective defect classification with fine-tuned ResNet18

Abstract

Industrial fruit inspection systems must operate reliably under dense multi-object interactions and continuous motion, yet most existing works evaluate detection or classification at the image level without ensuring temporal stability in video streams. We present a two-stage detection-tracking framework for stable multi-apple quality inspection in conveyor-belt environments. An orchard-trained YOLOv8 model performs apple localization, followed by ByteTrack multi-object tracking to maintain persistent identities. A ResNet18 defect classifier, fine-tuned on a healthy-defective fruit dataset, is applied to cropped apple regions. Track-level aggregation is introduced to enforce temporal consistency and reduce prediction oscillation across frames. We define video-level industrial metrics such as track-level defect ratio and temporal consistency to evaluate system robustness under realistic processing conditions. Results demonstrate improved stability compared to frame-wise inference, suggesting that integrating tracking is essential for practical automated fruit grading systems.