A PST Algorithm for FPs Suppression in Two-stage CNN Detection Methods

TL;DR

The paper tackles false positives in pedestrian detection by analyzing why two-stage CNN detectors misclassify proposals during training due to IoU-based labeling. It introduces Pedestrian-Sensitive Training (PST), a training-time mechanism that uses a lightweight pedestrian-sensitive classifier to re-evaluate and filter negative proposals, producing refined training targets Pt without increasing inference cost. Empirical results on CityPersons, CUHK-Occ, Caltech, and SY-Metro show that PST improves miss-rate performance and enables a compact detector, MetroNext-PST, to achieve competitive accuracy with favorable parameter and speed characteristics suitable for embedded edge devices. Overall, PST enables robust FP suppression in resource-constrained deployments and offers a practical path toward real-time pedestrian detection on mobile and embedded platforms.

Abstract

Pedestrian detection has been a hot spot in computer vision over the past decades due to the wide spectrum of promising applications, the major challenge of which is False Positives (FPs) that occur during pedestrian detection. The emergence various Convolutional Neural Network-based detection strategies substantially enhance the pedestrian detection accuracy but still not well solve this problem. This paper deeply analysis the detection framework of the two-stage CNN detection methods and find out false positives in detection results is due to its training strategy miss classify some false proposals, thus weakens the classification capability of following subnetwork and hardly to suppress false ones. To solve this problem, This paper proposes a pedestrian-sensitive training algorithm to effectively help two-stage CNN detection methods learn to distinguish the pedestrian and non-pedestrian samples and suppress the false positives in final detection results. The core of the proposed training algorithm is to redesign the training proposal generating pipeline of the two-stage CNN detection methods, which can avoid a certain number of false ones that mislead its training process. With the help of the proposed algorithm, the detection accuracy of the MetroNext, an smaller and accurate metro passenger detector, is further improved, which further decreases false ones in its metro passengers detection results. Based on various challenging benchmark datasets, experiment results have demonstrated that feasibility of the proposed algorithm to improve pedestrian detection accuracy by removing the false positives. Compared with the competitors, MetroNext-PST demonstrates better overall prediction performance in accuracy, total number of parameters, and inference time, thus it can become a practical solution for hunting pedestrian tailored for mobile and edge devices.

Figures (6)

• Figure 1: The training strategy of two-stage CNN-based pedestrian detection paradigm. Where "offsets & scrs" denotes the coordinate offsets and scores of predefined anchors. "bbs" represents the bounding boxes. "p" denotes the proposals, where foreground and background are represented by "fg,bg". "RoI Pooling" denotes the RoI Pooling layer. "$D_c(x),D_r(x)$" denote the classifier and regressor of the subnetwork. $\bigotimes$ denotes the operator for calculating IoU values.
• Figure 2: Some training samples are misclassified to negative ones using the IoU strategy.
• Figure 3: The processing pipeline of the PST algorithm, where $F(x)$ denotes the pedestrian sensitive classifier. $\phi_i$ represents the pedestrian confidence of the proposal. $\varepsilon$ represents the confidence threshold for pedestrians. All coloured dotted bounding boxes indicate these proposal are not used for training subnetwork.
• Figure 4: The architecture of the pedestrian-sensitive classifier. "Conv" denotes its convolutional block and the number in the bracket means the number of convolutional filters. "FC" stands for the Fully Connected layer.
• Figure 5: Comparison to the state-of-the-art detectors on benchmark datasets. (a) CUHK-Occ dataset. (b) Caltech dataset. (c) CityPersons dataset.