BBoxCut: A Targeted Data Augmentation Technique for Enhancing Wheat Head Detection Under Occlusions

2025-03-31

AI Summary

To address performance degradation in field-based wheat ear detection caused by leaf occlusion, inter-ear overlap, illumination variation, and motion blur, this paper proposes BBoxCut—a novel occlusion-aware data augmentation method tailored for wheat ear detection. Its core innovation lies in applying randomized local masking within ground-truth bounding boxes to faithfully simulate typical field occlusion patterns, rather than resorting to simplistic image-level cropping or padding. BBoxCut is detector-agnostic and seamlessly integrates with mainstream frameworks: it improves mean Average Precision (mAP) by 2.76, 3.26, and 1.9 percentage points on Faster R-CNN, FCOS, and DETR, respectively. Notably, detection accuracy for occluded wheat ears increases substantially, demonstrating superior robustness and generalization capability under complex agricultural conditions.

Technology Category

• Computer Vision: Object Detection & Categorization
• Machine Learning: Hardware-aware ML
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Wheat plays a critical role in global food security, making it one of the most extensively studied crops. Accurate identification and measurement of key characteristics of wheat heads are essential for breeders to select varieties for cross-breeding, with the goal of developing nutrient-dense, resilient, and sustainable cultivars. Traditionally, these measurements are performed manually, which is both time-consuming and inefficient. Advances in digital technologies have paved the way for automating this process. However, field conditions pose significant challenges, such as occlusions of leaves, overlapping wheat heads, varying lighting conditions, and motion blur. In this paper, we propose a novel data augmentation technique, BBoxCut, which uses random localized masking to simulate occlusions caused by leaves and neighboring wheat heads. We evaluated our approach using three state-of-the-art object detectors and observed mean average precision (mAP) gains of 2.76, 3.26, and 1.9 for Faster R-CNN, FCOS, and DETR, respectively. Our augmentation technique led to significant improvements both qualitatively and quantitatively. In particular, the improvements were particularly evident in scenarios involving occluded wheat heads, demonstrating the robustness of our method in challenging field conditions.

Problem

• Enhancing wheat head detection under occlusions in field conditions
• Addressing challenges like leaf occlusions and overlapping wheat heads
• Improving accuracy of automated wheat head measurement for breeding

Innovation

• BBoxCut uses localized masking for occlusion simulation
• Enhances wheat head detection under field conditions
• Improves mAP in three object detectors significantly