Zhor Benhafid
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhor Benhafid has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Sid-Ahmed Selouani
Laboratory of Research in Human-System Interaction (LARHSI), University of Moncton, Shippagan Campus, Moncton, Canada
6 shared publications - Abderrahmane Amrouche
Laboratoire de Communication Parlée et Traitement du Signal (LCPTS), Bab Ezzouar, Algeria
3 shared publications - Elise Mayrand
Secteur des sciences, Université de Moncton, Nouveau-Brunswick, Shippagan, Canada
2 shared publications - Mohammed Sidi Yakoub
Laboratoire de Recherche en Interaction Humain-Système (LARIHS), Université de Moncton, Shippagan, Canada
2 shared publications
Latest Publications
- Open Access
- 26 January 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (3) , 1236
Rarity-Aware Stratified Active Learning for Class-Imbalanced Industrial Object Detection
Object detection systems deployed in industrial environments are often constrained by limited annotation budgets, severe class imbalance, and heterogeneous visual conditions. Active learning (AL) aims to reduce labeling costs by selecting informative samples; however, existing strategies struggle to simultaneously ensure robust performance, rare-class coverage, and stability under realistic industrial constraints. In this work, we propose a rarity-aware, stratified AL framework for industrial object detection that explicitly aligns sample selection with class imbalance and annotation efficiency. The method relies on a composite image-level score that jointly captures model uncertainty, informativeness, and complementary diversity cues, while adaptively emphasizing rare classes. Crucially, a stratified querying mechanism is introduced to explicitly regulate class-wise sample allocation during selection, playing a key role in improving performance stability and rare-class coverage under severe imbalance, without sacrificing global informativeness. The proposed approach operates purely at the data-selection level, making it detector-agnostic and directly applicable to modern object detection pipelines. Experiments conducted on two real-world industrial datasets involving lobster and snow crab parts, using YOLOv10 and YOLOv12, demonstrate improved training stability and annotation efficiency across balanced, imbalanced, and noisy settings over multiple active learning cycles up to 15% labeled data. Complementary comparisons with fully supervised training further show that using only 45–65% of the labeled data is sufficient to retain more than 97% of full-supervision mAP@50 and over 90% of mAP@50:95.
- 26 January 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (3) , 1236
Open AccessRarity-Aware Stratified Active Learning for Class-Imbalanced Industrial Object Detection
Object detection systems deployed in industrial environments are often constrained by limited annotation budgets, severe class imbalance, and heterogeneous visual conditions. Active learning (AL) aims to reduce labeling costs by selecting informative samples; however, existing strategies struggle to simultaneously ensure robust performance, rare-class coverage, and stability under realistic industrial constraints. In this work, we propose a rarity-aware, stratified AL framework for industrial object detection that explicitly aligns sample selection with class imbalance and annotation efficiency. The method relies on a composite image-level score that jointly captures model uncertainty, informativeness, and complementary diversity cues, while adaptively emphasizing rare classes. Crucially, a stratified querying mechanism is introduced to explicitly regulate class-wise sample allocation during selection, playing a key role in improving performance stability and rare-class coverage under severe imbalance, without sacrificing global informativeness. The proposed approach operates purely at the data-selection level, making it detector-agnostic and directly applicable to modern object detection pipelines. Experiments conducted on two real-world industrial datasets involving lobster and snow crab parts, using YOLOv10 and YOLOv12, demonstrate improved training stability and annotation efficiency across balanced, imbalanced, and noisy settings over multiple active learning cycles up to 15% labeled data. Complementary comparisons with fully supervised training further show that using only 45–65% of the labeled data is sufficient to retain more than 97% of full-supervision mAP@50 and over 90% of mAP@50:95.
- Open Access
- 6 January 2026
- Published by Elsevier in Ocean Modelling
- Vol. 200
Explainable machine learning models for coastal pH forecasting at aquaculture-relevant thresholds in Eastern Canada
- 6 January 2026
- Published by Elsevier in Ocean Modelling
- Vol. 200
Open AccessExplainable machine learning models for coastal pH forecasting at aquaculture-relevant thresholds in Eastern Canada
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 130669-130690
SSRW-YOLO: Semi-Supervised Rarity-Weighted YOLO for Seafood Processing Applications
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 130669-130690
Open AccessSSRW-YOLO: Semi-Supervised Rarity-Weighted YOLO for Seafood Processing Applications
- 19 September 2024
- Published by Springer Nature in Circuits, Systems, and Signal Processing
- Vol. 44 (1) , 534-555
Attentive Context-Aware Deep Speaker Representations for Voice Biometrics in Adverse Conditions
- 19 September 2024
- Published by Springer Nature in Circuits, Systems, and Signal Processing
- Vol. 44 (1) , 534-555
Attentive Context-Aware Deep Speaker Representations for Voice Biometrics in Adverse Conditions
- 23 July 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83 (33) , 80261-80275
Deep normalization for light SpineNet speaker anti-spoofing systems
- 23 July 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83 (33) , 80261-80275
Deep normalization for light SpineNet speaker anti-spoofing systems
- 5 November 2023
- Published by Springer Nature in International Journal of Speech Technology
- Vol. 26 (4) , 881-894
Attention-based factorized TDNN for a noise-robust and spoof-aware speaker verification system
- 5 November 2023
- Published by Springer Nature in International Journal of Speech Technology
- Vol. 26 (4) , 881-894
Attention-based factorized TDNN for a noise-robust and spoof-aware speaker verification system
- Open Access
- 25 August 2023
- Published by Springer Nature in Anthropocene Coasts
- Vol. 6 (1) , 1-14
Spatiotemporal variability of pH in coastal waters of New Brunswick (Canada) and potential consequences for oyster aquaculture
There is a void in the knowledge of the acidification status of Eastern Canada's coastal waters. This knowledge is crucial to evaluating the threats posed to marine life, particularly oyster farming, a flagship of New Brunswick seafood production. In this study, we measured the temporal variability of pH and related environmental parameters in three bays of Northeastern New Brunswick. We also evaluated the potential impact of the observed pH levels on the Eastern oyster (Crassostrea virginica Gmelin, 1791), based on the available literature on this species’ sensitivity to acidification. We investigated the presence of inherent cycles of pH with the Fourier transform and the spectral filtering technique. Our results show that pH is highly variable in the studied area, with values ranging from 7.31 to 8.90. A seasonal effect was apparent, as the pH fluctuations were set at the lowest level in winter when the cover of ice and snow on the bay was present. The spectral analysis revealed a clear semidiurnal tidal pattern of pH, this variable being inversely related to the water level in summer and directly related to it in winter. The spectral subtraction of all the tidal components allowed the detection of a circadian rhythm that was not in pace with the alternation between day and night but rather slowly drifted so that the pH troughs occurred at night during the full moon period. Short periodicities of circa 8 and 6 h also existed in two of the three bays. Based on current knowledge of C. virginica’s sensitivity to acidification, this species’ recruitment, growth, and survival are unlikely to be impacted by the present pH levels in the studied area. However, further acidification might overcome the resilience of C. virginica, especially that of the larvae that are produced during the winter in commercial hatcheries.
- 25 August 2023
- Published by Springer Nature in Anthropocene Coasts
- Vol. 6 (1) , 1-14
Open AccessSpatiotemporal variability of pH in coastal waters of New Brunswick (Canada) and potential consequences for oyster aquaculture
There is a void in the knowledge of the acidification status of Eastern Canada's coastal waters. This knowledge is crucial to evaluating the threats posed to marine life, particularly oyster farming, a flagship of New Brunswick seafood production. In this study, we measured the temporal variability of pH and related environmental parameters in three bays of Northeastern New Brunswick. We also evaluated the potential impact of the observed pH levels on the Eastern oyster (Crassostrea virginica Gmelin, 1791), based on the available literature on this species’ sensitivity to acidification. We investigated the presence of inherent cycles of pH with the Fourier transform and the spectral filtering technique. Our results show that pH is highly variable in the studied area, with values ranging from 7.31 to 8.90. A seasonal effect was apparent, as the pH fluctuations were set at the lowest level in winter when the cover of ice and snow on the bay was present. The spectral analysis revealed a clear semidiurnal tidal pattern of pH, this variable being inversely related to the water level in summer and directly related to it in winter. The spectral subtraction of all the tidal components allowed the detection of a circadian rhythm that was not in pace with the alternation between day and night but rather slowly drifted so that the pH troughs occurred at night during the full moon period. Short periodicities of circa 8 and 6 h also existed in two of the three bays. Based on current knowledge of C. virginica’s sensitivity to acidification, this species’ recruitment, growth, and survival are unlikely to be impacted by the present pH levels in the studied area. However, further acidification might overcome the resilience of C. virginica, especially that of the larvae that are produced during the winter in commercial hatcheries.