Salvato in:
Dettagli Bibliografici
Autori principali: Golizadeh, Mahdi, Golizadeh, Nassibeh, Keyvanrad, Mohammad Ali, Shirazi, Hossein
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2508.03317
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916882393595904
author Golizadeh, Mahdi
Golizadeh, Nassibeh
Keyvanrad, Mohammad Ali
Shirazi, Hossein
author_facet Golizadeh, Mahdi
Golizadeh, Nassibeh
Keyvanrad, Mohammad Ali
Shirazi, Hossein
contents Object detection has achieved remarkable accuracy through deep learning, yet these improvements often come with increased computational cost, limiting deployment on resource-constrained devices. Knowledge Distillation (KD) provides an effective solution by enabling compact student models to learn from larger teacher models. However, adapting KD to object detection poses unique challenges due to its dual objectives-classification and localization-as well as foreground-background imbalance and multi-scale feature representation. This review introduces a novel architecture-centric taxonomy for KD methods, distinguishing between CNN-based detectors (covering backbone-level, neck-level, head-level, and RPN/RoI-level distillation) and Transformer-based detectors (including query-level, feature-level, and logit-level distillation). We further evaluate representative methods using the MS COCO and PASCAL VOC datasets with mAP@0.5 as performance metric, providing a comparative analysis of their effectiveness. The proposed taxonomy and analysis aim to clarify the evolving landscape of KD in object detection, highlight current challenges, and guide future research toward efficient and scalable detection systems.
format Preprint
id arxiv_https___arxiv_org_abs_2508_03317
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Architectural Insights into Knowledge Distillation for Object Detection: A Comprehensive Review
Golizadeh, Mahdi
Golizadeh, Nassibeh
Keyvanrad, Mohammad Ali
Shirazi, Hossein
Computer Vision and Pattern Recognition
68T07
I.4.8
Object detection has achieved remarkable accuracy through deep learning, yet these improvements often come with increased computational cost, limiting deployment on resource-constrained devices. Knowledge Distillation (KD) provides an effective solution by enabling compact student models to learn from larger teacher models. However, adapting KD to object detection poses unique challenges due to its dual objectives-classification and localization-as well as foreground-background imbalance and multi-scale feature representation. This review introduces a novel architecture-centric taxonomy for KD methods, distinguishing between CNN-based detectors (covering backbone-level, neck-level, head-level, and RPN/RoI-level distillation) and Transformer-based detectors (including query-level, feature-level, and logit-level distillation). We further evaluate representative methods using the MS COCO and PASCAL VOC datasets with mAP@0.5 as performance metric, providing a comparative analysis of their effectiveness. The proposed taxonomy and analysis aim to clarify the evolving landscape of KD in object detection, highlight current challenges, and guide future research toward efficient and scalable detection systems.
title Architectural Insights into Knowledge Distillation for Object Detection: A Comprehensive Review
topic Computer Vision and Pattern Recognition
68T07
I.4.8
url https://arxiv.org/abs/2508.03317