Type of Publication

Journal Articles

Date:

5 /

2022

Status

Published

DOI:

10.1109/ACCESS.2022.3175195

Reducing Overconfidence Predictions in Autonomous Driving Perception

Featured in:

IEEE Access, vol. 10, pp. 54805-54821, 2022

Authors:

Gledson Melotti, Cristiano Premebida, Jordan J. Bird, Diego R. Faria and Nuno Gonçalves

Abstract

In state-of-the-art deep learning for object recognition, Softmax and Sigmoid layers are most commonly employed as the predictor outputs. Such layers often produce overconfidence predictions rather than proper probabilistic scores, which can thus harm the decision-making of ‘critical’ perception systems applied in autonomous driving and robotics. Given this, we propose a probabilistic approach based on distributions calculated out of the Logit layer scores of pre-trained networks which are then used to constitute new decision layers based on Maximum Likelihood ( ML ) and Maximum a-Posteriori (MAP) inference. We demonstrate that the hereafter called ML and MAP layers are more suitable for probabilistic interpretations than Softmax and Sigmoid-based predictions for object recognition. We explore distinct sensor modalities via RGB images and LiDARs (RV: range-view) data from the KITTI and Lyft Level-5 datasets, where our approach shows promising performance compared to the usual Softmax and Sigmoid layers, with the benefit of enabling interpretable probabilistic predictions. Another advantage of the approach introduced in this paper is that the so-called ML and MAP layers can be implemented in existing trained networks, that is, the approach benefits from the output of the Logit layer of pre-trained networks. Thus, there is no need to carry out a new training phase since the ML and MAP layers are used in the test/prediction phase. The Classification results are presented using reliability diagrams, while detection results are illustrated using precision-recall curves.

Citation
Gledson Melotti, Cristiano Premebida, Jordan J. Bird, Diego R. Faria and Nuno Gonçalves (2022). Reducing overconfidence predictions in autonomous driving perception. IEEE Access, 10, 54805-54821. DOI: 10.1109/ACCESS.2022.3175195

Related Content

Researcher Coordinator, VIS TEAM Leader
PhD Student
No tagged content to show
No tagged content to show
No tagged content to show

RECENT PUBLICATIONS

Graph-Based Radiomics Feature Extraction from 2D Retina Images

Authors: Ofélio Jorreia; Nuno Gonçalves; Rui Cortesão
Featured in: IEEE Access

Book of Extended Abstracts of the 12th Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA 2025)

Authors: Aitana Menárguez-Box; Angel Navarro; Antonio Requena Jiménez; Brenda Nogueira; Dylan Perdigão; Farhad Shadmand; Iurii Medvedev; Marco Alexandre Tomás Tereso; Maria del Mar Coch-Alcina; Matheus Kovaleski; Miguel Leão; Teresa M.C. Pereira; Tomás Silva Santos Rocha
Featured in: 12th Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA 2025)

How to Rethink Education with Artificial Intelligence: The Portuguese Use Case from Political and Practical Perspectives

Authors: Nuno Gonçalves; Maria Helena Monteiro
Featured in: Futures'2025 Conference Artificial Intelligence in Education

suggested news

Four papers presented @ IbPRIA 2025
Prof. Nuno participates in Conference on Digital Governance
ISR-UC maintains the “Excellent” rating in FCT evaluation!

RECENT PROJECTS

FACING2 – Face Image Understanding
VISUAL-ID – Unique Visual Identities in Graphics, Images and Faces
UniqueMark

Institute of Systems and Robotics Department of Electrical and Computers Engineering University of Coimbra