Selected work

Representative Papers

Domain-Aware Lightweight Spectral-Grouped Convolutions for Hyperspectral Fish Freshness Classification

Aug 12, 2026

This study addresses the challenge of accurately classifying fish freshness from hyperspectral images, which is hindered by spectral dominance, ordinal label structure, and severe sample scarcity that limit the effectiveness of conventional deep learning approaches. To overcome these limitations, this work proposes SGNet, a lightweight network that introduces domain-aware spectral grouping convolution to decouple spectral and spatial features. By integrating depthwise separable convolution with a dual channel–spatial attention mechanism, SGNet enhances discriminative capability with minimal computational overhead. Evaluated on a self-collected dataset of salmon samples stored over 16 days, SGNet achieves a classification accuracy of 97.8% and an average absolute error of 0.64 days using only 4.75 million parameters—reducing parameter count by 5–18× compared to ResNet-50 and Vision Transformer while maintaining high accuracy and real-time performance.

0 citationsRead paper

Few-Shot Ordinal Learning for Day-Wise Freshness Estimation with Hyperspectral Fish Images

Aug 12, 2026

This study addresses the challenges of scarce annotations and high inter-individual variability in daily-level freshness estimation of fish from hyperspectral images by introducing few-shot learning to food quality assessment for the first time. Each fish fillet is modeled as an independent few-shot ordinal regression task. The proposed method integrates a cumulative threshold ordinal prediction head with a CORAL-style regression architecture, enhanced by biologically inspired monotonicity constraints and embedding smoothness regularization to ensure temporally coherent freshness predictions. Under a strict evaluation protocol involving unseen fish fillets, the approach achieves a mean absolute error of 1.58 days and a within-two-days accuracy of 72.3% using only three labeled days per fillet, significantly outperforming scalar regression and label distribution baselines.

0 citationsRead paper

Concurrent validity of computer-vision artificial intelligence player tracking software using broadcast footage

Aug 26, 2025

Commercial AI-based player tracking systems lack systematic validation for concurrent validity in broadcast video, particularly regarding positional accuracy, velocity estimation, and total distance covered during elite football matches. Method: This study conducts the first comprehensive concurrent validity assessment of three leading commercial AI tracking solutions on FIFA World Cup broadcast footage, using multi-camera, high-definition TRACAB Gen 5 data as the ground-truth reference. It quantifies errors in position (RMSE), speed, and cumulative running distance, while analyzing the impact of camera viewpoint (e.g.,俯角, wide-angle) and resolution. Results: Positional RMSE ranges from 1.68–16.39 m; speed error from 0.34–2.38 m/s; and total distance bias from −21.8% to +24.3%. Crucially, tactical camera angles—especially elevated and wide-field views—significantly improve localization accuracy, demonstrating that broadcast geometry fundamentally constrains AI tracking performance. This work establishes the first empirical benchmark and methodological framework for validating AI-driven sports analytics in live broadcast environments.

0 citationsRead paper

Recent publications

Latest Papers

Domain-Aware Lightweight Spectral-Grouped Convolutions for Hyperspectral Fish Freshness Classification

Aug 12, 2026

This study addresses the challenge of accurately classifying fish freshness from hyperspectral images, which is hindered by spectral dominance, ordinal label structure, and severe sample scarcity that limit the effectiveness of conventional deep learning approaches. To overcome these limitations, this work proposes SGNet, a lightweight network that introduces domain-aware spectral grouping convolution to decouple spectral and spatial features. By integrating depthwise separable convolution with a dual channel–spatial attention mechanism, SGNet enhances discriminative capability with minimal computational overhead. Evaluated on a self-collected dataset of salmon samples stored over 16 days, SGNet achieves a classification accuracy of 97.8% and an average absolute error of 0.64 days using only 4.75 million parameters—reducing parameter count by 5–18× compared to ResNet-50 and Vision Transformer while maintaining high accuracy and real-time performance.

0 citationsRead paper

Few-Shot Ordinal Learning for Day-Wise Freshness Estimation with Hyperspectral Fish Images

Aug 12, 2026

This study addresses the challenges of scarce annotations and high inter-individual variability in daily-level freshness estimation of fish from hyperspectral images by introducing few-shot learning to food quality assessment for the first time. Each fish fillet is modeled as an independent few-shot ordinal regression task. The proposed method integrates a cumulative threshold ordinal prediction head with a CORAL-style regression architecture, enhanced by biologically inspired monotonicity constraints and embedding smoothness regularization to ensure temporally coherent freshness predictions. Under a strict evaluation protocol involving unseen fish fillets, the approach achieves a mean absolute error of 1.58 days and a within-two-days accuracy of 72.3% using only three labeled days per fillet, significantly outperforming scalar regression and label distribution baselines.

0 citationsRead paper

Concurrent validity of computer-vision artificial intelligence player tracking software using broadcast footage

Aug 26, 2025

Commercial AI-based player tracking systems lack systematic validation for concurrent validity in broadcast video, particularly regarding positional accuracy, velocity estimation, and total distance covered during elite football matches. Method: This study conducts the first comprehensive concurrent validity assessment of three leading commercial AI tracking solutions on FIFA World Cup broadcast footage, using multi-camera, high-definition TRACAB Gen 5 data as the ground-truth reference. It quantifies errors in position (RMSE), speed, and cumulative running distance, while analyzing the impact of camera viewpoint (e.g.,俯角, wide-angle) and resolution. Results: Positional RMSE ranges from 1.68–16.39 m; speed error from 0.34–2.38 m/s; and total distance bias from −21.8% to +24.3%. Crucially, tactical camera angles—especially elevated and wide-field views—significantly improve localization accuracy, demonstrating that broadcast geometry fundamentally constrains AI tracking performance. This work establishes the first empirical benchmark and methodological framework for validating AI-driven sports analytics in live broadcast environments.

0 citationsRead paper