Introspective classification for pedestrian detection

Abstract

State-of-the-art pedestrian detectors are capable of finding humans in images with reasonable accuracy. However, accurate object detectors such as Integral Channel Features (ICF) do not provide good reliability; they are unable to identify detections which they are less confident (or more uncertain) about. We apply existing methods for generating probabilistic measures from classifier scores (such as Platt exponential scaling and Isotonic Regression) and compare these to Gaussian Process classifiers (GPCs), which can provide more informative predictive variance. GPCs are less accurate than ICF classifiers, but GPCs and Adaboost with Platt scaling both provide improved reliability over existing methods.

Introspection of DNN-Based Perception Functions in Automated Driving Systems: State-of-the-Art and Open Research Challenges

“…Then such methods can query to introspect the decisions of the classification model. Confidence Calibration [37], [48], [49], [50], [51], [52], [53], [54] Use of confidence outputs, or logits from neural network to process them further.…”

“…In [48], confidence score processing methods, non-parametrised sigmoid, Platt scaling [49], isotonic regression, and Gaussian processes are tested with different classifiers to see their performance on classifying pedestrians in detected objects. The study aims to compare methods and provide which achieves realistic confidence scores for introspection.…”

Estimation of fault probability in medium voltage feeders through calibration techniques in classification models

“…In Gosztolya and Busa-Fekete (2018), calibration methods such as Linear Scaling, Platt scaling and Isotonic Regression are applied to the AdaBoost.MH algorithm (Schapire and Singer 1999) for a speech recognition task. In Blair et al (2014) calibration is applied as postprocessing step for an object detection task. The authors investigate the performance of Platt scaling and Isotonic Regression in converting a score, obtained from a machine learning detector based on an Adaboost classifier or an SVM, to a probability representing confidence in measuring the presence of an object in a given location.…”

CogSense: A Cognitively Inspired Framework for Perception Adaptation

“…Aleatoric uncertainty is the uncertainty of our input, and epistemic uncertainty is the uncertainty of our model. Some methods to estimate epistemic uncertainty are Bayesian Neural Networks [18], ensemble methods [19,20], Monte Carlo dropout [21], sampling-free methods [22], or directly from the input [23]. Uncertainty can be used as an indicator of the likelihood of an error.…”