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

2026-02-22

AI Summary

This study addresses the challenge of reliable multi-apple quality assessment in dense conveyor belt scenarios, where existing fruit detection methods suffer from insufficient temporal stability. To overcome this limitation, the authors propose a two-stage detection-tracking framework: apples are first localized using YOLOv8, followed by identity maintenance via ByteTrack, and defect classification is performed on cropped regions using ResNet18. Crucially, trajectory-level prediction aggregation is introduced to significantly enhance temporal consistency. This work represents the first application of multi-object tracking combined with trajectory-level aggregation to apple quality inspection, establishing novel industrial-oriented evaluation metrics—trajectory-level defect rate and temporal consistency—and demonstrating the necessity and effectiveness of tracking mechanisms for automated fruit grading systems.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Security and Privacy: Tracking, profiling, and countermeasures against them
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

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.

Problem

• temporal stability
• multi-object interaction
• fruit inspection
• conveyor-belt environment
• video-level consistency

Innovation

• detection-tracking framework
• temporal consistency
• multi-object tracking
• fruit quality inspection
• track-level aggregation