Mitigating Cross-Dataset Feature Shifts via a Multi-Loss Self-Reconstruction Filter: A Correction-Detection Framework for UAV-Based Yuzu Detection
Preprint
- 25 May 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
Deep learning-based yuzu detection systems often suffer performance degradation when the training and inference datasets differ in collection sites, acquisition years, and imaging conditions. To address this issue, this study proposes a multi-loss self-reconstruction (MLSR) filter and an MLSR-based correction-detection framework.The proposed MLSR filter is an encoder-decoder architecture, where the input image is encoded into feature representations and then decoded back into the image space. It aims to reconstruct natural image characteristics while suppressing detection-irrelevant variations. To achieve this objective, we propose a staged optimization strategy with multi-loss constraints to train the filter, in which L1 and SSIM losses constrain the reconstruction to recover the basic image structure, the VGG perceptual loss constrains it to preserve detection-relevant features, and the PatchGAN adversarial loss prevents the reconstructed images from deviating excessively from natural image characteristics. In the correction-detection framework, the trained MLSR filter is applied to both training and inference datasets before yuzu detection. The proposed correction-detection framework was evaluated on UAV (Unmanned Aerial Vehicle) imagery collected under different orchards, years, and illumination conditions. The results showed consistent improvements for YOLOv5, YOLOv8, and YOLOv11, with maximum gains of 5.62% in F1-score, 7.72% in mAP50, and 6.95% in mAP50-90. We argued that MLSR filtering mitigated feature shifts between the training and inference datasets, allowing YOLO to capture more consistent patterns across them and thereby improving detection performance. However, this mitigation was reflected in YOLO backbones but did not introduce obvious visual changes. Therefore, we measured the similarity between the training and inference datasets based on YOLO backbone representations before and after MLSR filtering. The results showed that YOLO perceived higher feature similarity between the datasets at detection-relevant scales after MLSR filtering, consistent with the reported detection improvements.Keywords
This publication has 19 references indexed in Scilit:
- Data generation using Pix2Pix to improve YOLO v8 performance in UAV-based Yuzu detectionSmart Agricultural Technology, 2025
- Agricultural object detection with You Only Look Once (YOLO) Algorithm: A bibliometric and systematic literature reviewComputers and Electronics in Agriculture, 2024
- Agroview: Cloud-based application to process, analyze and visualize UAV-collected data for precision agriculture applications utilizing artificial intelligenceComputers and Electronics in Agriculture, 2020
- Cosine similarity measures for intuitionistic fuzzy sets and their applicationsMathematical and Computer Modelling, 2011
- On Early Stopping in Gradient Descent LearningConstructive Approximation, 2007
- Best practices for convolutional neural networks applied to visual document analysis2005
- Color transfer between imagesIEEE Computer Graphics and Applications, 2001
- A multiscale retinex for bridging the gap between color images and the human observation of scenesIEEE Transactions on Image Processing, 1997
- Adaptive histogram equalization and its variationsComputer Vision, Graphics, and Image Processing, 1987
- Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in positionBiological Cybernetics, 1980