Inlier-Centric Post-Training Quantization for Object Detection Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kim, Minsu, Lee, Dongyeun, Yu, Jaemyung, Hur, Jiwan, Kim, Giseop, Kim, Junmo
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917244626272256
author Kim, Minsu
Lee, Dongyeun
Yu, Jaemyung
Hur, Jiwan
Kim, Giseop
Kim, Junmo
author_facet Kim, Minsu
Lee, Dongyeun
Yu, Jaemyung
Hur, Jiwan
Kim, Giseop
Kim, Junmo
contents Object detection is pivotal in computer vision, yet its immense computational demands make deployment slow and power-hungry, motivating quantization. However, task-irrelevant morphologies such as background clutter and sensor noise induce redundant activations (or anomalies). These anomalies expand activation ranges and skew activation distributions toward task-irrelevant responses, complicating bit allocation and weakening the preservation of informative features. Without a clear criterion to distinguish anomalies, suppressing them can inadvertently discard useful information. To address this, we present InlierQ, an inlier-centric post-training quantization approach that separates anomalies from informative inliers. InlierQ computes gradient-aware volume saliency scores, classifies each volume as an inlier or anomaly, and fits a posterior distribution over these scores using the Expectation-Maximization (EM) algorithm. This design suppresses anomalies while preserving informative features. InlierQ is label-free, drop-in, and requires only 64 calibration samples. Experiments on the COCO and nuScenes benchmarks show consistent reductions in quantization error for camera-based (2D and 3D) and LiDAR-based (3D) object detection.
format Preprint
id arxiv_https___arxiv_org_abs_2602_03472
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Inlier-Centric Post-Training Quantization for Object Detection Models
Kim, Minsu
Lee, Dongyeun
Yu, Jaemyung
Hur, Jiwan
Kim, Giseop
Kim, Junmo
Computer Vision and Pattern Recognition
Object detection is pivotal in computer vision, yet its immense computational demands make deployment slow and power-hungry, motivating quantization. However, task-irrelevant morphologies such as background clutter and sensor noise induce redundant activations (or anomalies). These anomalies expand activation ranges and skew activation distributions toward task-irrelevant responses, complicating bit allocation and weakening the preservation of informative features. Without a clear criterion to distinguish anomalies, suppressing them can inadvertently discard useful information. To address this, we present InlierQ, an inlier-centric post-training quantization approach that separates anomalies from informative inliers. InlierQ computes gradient-aware volume saliency scores, classifies each volume as an inlier or anomaly, and fits a posterior distribution over these scores using the Expectation-Maximization (EM) algorithm. This design suppresses anomalies while preserving informative features. InlierQ is label-free, drop-in, and requires only 64 calibration samples. Experiments on the COCO and nuScenes benchmarks show consistent reductions in quantization error for camera-based (2D and 3D) and LiDAR-based (3D) object detection.
title Inlier-Centric Post-Training Quantization for Object Detection Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.03472