Event-Based Crossing Dataset (EBCD)

TL;DR

The EBCD is introduced, a comprehensive dataset tailored for pedestrian and vehicle detection in dynamic outdoor environments, incorporating a multi-thresholding framework to refine event representations, and is presented as publicly available to propel further advancements in low-latency, high-fidelity neuromorphic imaging.

Abstract

Event-based vision revolutionizes traditional image sensing by capturing asynchronous intensity variations rather than static frames, enabling ultrafast temporal resolution, sparse data encoding, and enhanced motion perception. While this paradigm offers significant advantages, conventional event-based datasets impose a fixed thresholding constraint to determine pixel activations, severely limiting adaptability to real-world environmental fluctuations. Lower thresholds retain finer details but introduce pervasive noise, whereas higher thresholds suppress extraneous activations at the expense of crucial object information. To mitigate these constraints, we introduce the Event-Based Crossing Dataset (EBCD), a comprehensive dataset tailored for pedestrian and vehicle detection in dynamic outdoor environments, incorporating a multi-thresholding framework to refine event representations. By capturing event-based images at ten distinct threshold levels (4, 8, 12, 16, 20, 30, 40, 50, 60, and 75), this dataset facilitates an extensive assessment of object detection performance under varying conditions of sparsity and noise suppression. We benchmark state-of-the-art detection architectures-including YOLOv4, YOLOv7, EfficientDet-b0, MobileNet-v1, and Histogram of Oriented Gradients (HOG)-to experiment upon the nuanced impact of threshold selection on detection performance. By offering a systematic approach to threshold variation, we foresee that EBCD fosters a more adaptive evaluation of event-based object detection, aligning diverse neuromorphic vision with real-world scene dynamics. We present the dataset as publicly available to propel further advancements in low-latency, high-fidelity neuromorphic imaging: https://ieee-dataport.org/documents/event-based-crossing-dataset-ebcd

Figures (6)

• Figure 1: Dataset directories and files organization
• Figure 2: Comparison of object detection model inferences across varying threshold levels. The original image contains four objects: two pedestrians and two vehicles. The first row displays the baseline ground truth (yellow) bounding boxes for reference. As $T_h$ increases, detections become more selective, reducing false positives while incorporating less noise.
• Figure 3: Stacked pixel activations for $T_4$ (a), $T_{20}$ (b), $T_{50}$ (c), and $T_{75}$ (d), illustrating noise suppression with increasing thresholds. At $T_4$, extensive pixel activations obscure object boundaries, with 90.09% of activations occurring outside bounding boxes. As the threshold increases, background noise diminishes, reducing extraneous activations to 14.24%, 6.62%, and 5.96% at $T_{20}$, $T_{50}$, and $T_{75}$, respectively, improving object visibility and segmentation.
• Figure 4: Pixel activation distributions for an outdoor event-based image (EBCD, blue) and an indoor event-based image (SEFD, red) at $T_4$. The outdoor image exhibits significantly higher and more dispersed activations due to environmental noise, whereas the indoor image presents a more localized and controlled activation pattern.
• Figure 5: The graphs present the AP50 (a) and AP75 (b) across different decision thresholds object detection models: YOLO-v4, YOLO-v7, EfficientDet-b0, MobileNet-v1, and HOG. The yellow vertical highlight indicates the optimal threshold ($T_{16}$) that yields the best average performance across models.