Multivariate Confidence Calibration for Object Detection (2020-04-01T00:00:00.000000Z)

TL;DR

This work presents a novel framework to measure and calibrate biased (or miscalibrated) confidence estimates of object detection methods and proposes a new measure to evaluate miscalibration of object detectors.

Abstract

Unbiased confidence estimates of neural networks are crucial especially for safety-critical applications. Many methods have been developed to calibrate biased confidence estimates. Though there is a variety of methods for classification, the field of object detection has not been addressed yet. Therefore, we present a novel framework to measure and calibrate biased (or miscalibrated) confidence estimates of object detection methods 1. The main difference to related work in the field of classifier calibration is that we also use additional information of the regression output of an object detector for calibration. Our approach allows, for the first time, to obtain calibrated confidence estimates with respect to image location and box scale. In addition, we propose a new measure to evaluate miscalibration of object detectors. Finally, we show that our developed methods outperform state-of-the-art calibration models for the task of object detection and provides reliable confidence estimates across different locations and scales.

Authors

Fabian Küppers

1 papers

Jan Kronenberger

1 papers

Amirhossein Shantia

1 papers

Multivariate Confidence Calibration for Object Detection

TL;DR

Abstract

Authors

References26 items

Uncertainty Estimation in One-Stage Object Detection

Feature Selective Anchor-Free Module for Single-Shot Object Detection

Generalized Intersection Over Union: A Metric and a Loss for Bounding Box Regression

Calibrating Uncertainties in Object Localization Task

Relaxed Softmax: Efficient Confidence Auto-Calibration for Safe Pedestrian Detection

Learning for Single-Shot Confidence Calibration in Deep Neural Networks Through Stochastic Inferences

Acquisition of Localization Confidence for Accurate Object Detection

Accurate Uncertainties for Deep Learning Using Calibrated Regression

On Calibration of Modern Neural Networks

Beta calibration: a well-founded and easily implemented improvement on logistic calibration for binary classifiers

What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?

Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors

R-FCN: Object Detection via Region-based Fully Convolutional Networks

SSD: Single Shot MultiBox Detector

Binary Classifier Calibration Using an Ensemble of Near Isotonic Regression Models

Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Obtaining Well Calibrated Probabilities Using Bayesian Binning

Constructions for a bivariate beta distribution

Microsoft COCO: Common Objects in Context

Transforming classifier scores into accurate multiclass probability estimates

Obtaining calibrated probability estimates from decision trees and naive Bayesian classifiers

On Discriminative vs. Generative Classifiers: A comparison of logistic regression and naive Bayes

Multivariate Generalized Beta Distributions with Applications to Utility Assessment

Learning for Single-Shot Confidence Calibration in Deep Neural Networks through Stochastic Inferences

Probabilistic Outputs for Support vector Machines and Comparisons to Regularized Likelihood Methods

Field of Study

Journal Information

Name

Page