Comprehensive Parameter Sweep for Learning-Based Detector on Traffic Lights

Abstract

Abstract. Determining the optimal parameters for a given detection algorithm is not straightforward and what ends up as the final values is mostly based on experience and heuristics. In this paper we investigate the influence of three basic parameters in the widely used Aggregate Channel Features (ACF) object detector applied for traffic light detection. Additionally, we perform an exhaustive search for the optimal parameters for the night time data from the LISA Traffic Light Dataset. The optimized detector reaches an Area-Under-Curve of 66.63 % on calculated precision-recall curve.

Evaluating State-of-the-Art Object Detector on Challenging Traffic Light Data

“…As for most other computer vision research areas, the popular combination of using Histogram of Oriented Gradients features together with a SVM classifier was introduced in [2]. The learning-based Aggregated Channel Features (ACF) detector have seen a large use in TLD, and have shown superior performance over the heuristic models both during day and night time [10,9]. TLD using Convolutional Neural Network (CNN) is introduced in [13,12], where a CNN is used detects and recognize the traffic lights using region-of-interest information provided by an onboard GPS sensor.…”

“…DAS applications are already widely implemented in newer vehicles, such as emergency breaking, automatic lane changing, keeping the advertised speed limit, and adaptive cruise control. DAS applications can usually be split into looking-in [28], such as hands activity recognition [19] and looking-out applications, such as detection of other vehicles, pedestrians [5], traffic signs [18] or traffic lights [9]. In 2012, 683 people died and 133,000 people were injured in crashes related to red light running in the USA [26], making traffic light detection a vital part of both self-driving cars and DAS.…”