TY - GEN
T1 - Q-YOLO
T2 - 7th Asian Conference on Pattern Recognition, ACPR 2023
AU - Wang, Mingze
AU - Sun, Huixin
AU - Shi, Jun
AU - Liu, Xuhui
AU - Cao, Xianbin
AU - Zhang, Luping
AU - Zhang, Baochang
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.
PY - 2023
Y1 - 2023
N2 - Real-time object detection plays a vital role in various computer vision applications. However, deploying real-time object detectors on resource-constrained platforms poses challenges due to high computational and memory requirements. This paper describes a low-bit quantization method to build a highly efficient one-stage detector, dubbed as Q-YOLO, which can effectively address the performance degradation problem caused by activation distribution imbalance in traditional quantized YOLO models. Q-YOLO introduces a fully end-to-end Post-Training Quantization (PTQ) pipeline with a well-designed Unilateral Histogram-based (UH) activation quantization scheme, which determines the maximum truncation values through histogram analysis by minimizing the Mean Squared Error (MSE) quantization errors. Extensive experiments on the COCO dataset demonstrate the effectiveness of Q-YOLO, outperforming other PTQ methods while achieving a more favorable balance between accuracy and computational cost. This research contributes to advancing the efficient deployment of object detection models on resource-limited edge devices, enabling real-time detection with reduced computational and memory overhead.
AB - Real-time object detection plays a vital role in various computer vision applications. However, deploying real-time object detectors on resource-constrained platforms poses challenges due to high computational and memory requirements. This paper describes a low-bit quantization method to build a highly efficient one-stage detector, dubbed as Q-YOLO, which can effectively address the performance degradation problem caused by activation distribution imbalance in traditional quantized YOLO models. Q-YOLO introduces a fully end-to-end Post-Training Quantization (PTQ) pipeline with a well-designed Unilateral Histogram-based (UH) activation quantization scheme, which determines the maximum truncation values through histogram analysis by minimizing the Mean Squared Error (MSE) quantization errors. Extensive experiments on the COCO dataset demonstrate the effectiveness of Q-YOLO, outperforming other PTQ methods while achieving a more favorable balance between accuracy and computational cost. This research contributes to advancing the efficient deployment of object detection models on resource-limited edge devices, enabling real-time detection with reduced computational and memory overhead.
KW - Post-training Quantization
KW - Real-time Object Detection
UR - https://www.scopus.com/pages/publications/85177420289
U2 - 10.1007/978-3-031-47665-5_25
DO - 10.1007/978-3-031-47665-5_25
M3 - 会议稿件
AN - SCOPUS:85177420289
SN - 9783031476648
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 307
EP - 321
BT - Pattern Recognition - 7th Asian Conference, ACPR 2023, Proceedings
A2 - Lu, Huimin
A2 - Blumenstein, Michael
A2 - Cho, Sung-Bae
A2 - Liu, Cheng-Lin
A2 - Yagi, Yasushi
A2 - Kamiya, Tohru
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 5 November 2023 through 8 November 2023
ER -