Search

For ordinary IR-illuminators in nighttime surveillance system, insufficient illumination may cause misdetection for faraway object while excessive illumination leads to over-exposure of nearby object. To overcome these two problems, we use the MI3 image dataset, which is established by multi-intensity IR-illumination (MIIR), as our benchmark dataset for modern object detection methods. We first provide complete annotations for the MI3 as its current ground-truth is incomplete. Then, we use these multi-intensity illuminated IR videos to evaluate several widely used object detectors, i.e., SSD, YOLO, Faster R-CNN, and Mask R-CNN, by analyzing the effective range of different illumination intensities. By including a tracking scheme, as well as developing of a new fusion method for different illumination intensities to improve the performance, the proposed approach may serve as a new benchmark of face and object detection for a wide range of distances. The new dataset (Dataset is available: https://ieee-dataport.org/documents/mi3) with more complete annotations and source codes (Codes are available: https://github.com/thesuperorange/deepMI3) is available online.

Featured Application: This paper proposes a deep-learning-based vehicle detection technique that can achieve effective detection performance under extreme illumination conditions. The technique can be used in on-road driver assistance tools and autonomous vehicles. Most object detection models cannot achieve satisfactory performance under nighttime and other insufficient illumination conditions, which may be due to the collection of data sets and typical labeling conventions. Public data sets collected for object detection are usually photographed with sufficient ambient lighting. However, their labeling conventions typically focus on clear objects and ignore blurry and occluded objects. Consequently, the detection performance levels of traditional vehicle detection techniques are limited in nighttime environments without sufficient illumination. When objects occupy a small number of pixels and the existence of crucial features is infrequent, traditional convolutional neural networks (CNNs) may suffer from serious information loss due to the fixed number of convolutional operations. This study presents solutions for data collection and the labeling convention of nighttime data to handle various types of situations, including in-vehicle detection. Moreover, the study proposes a specifically optimized system based on the Faster region-based CNN model. The system has a processing speed of 16 frames per second for 500 × 375-pixel images, and it achieved a mean average precision (mAP) of 0.8497 in our validation segment involving urban nighttime and extremely inadequate lighting conditions. The experimental results demonstrated that our proposed methods can achieve high detection performance in various nighttime environments, such as urban nighttime conditions with insufficient illumination, and extremely dark conditions with nearly no lighting. The proposed system outperforms original methods that have an mAP value of approximately 0.2. [ABSTRACT FROM AUTHOR]

Most object detection models cannot achieve satisfactory performance under nighttime and other insufficient illumination conditions, which may be due to the collection of data sets and typical labeling conventions. Public data sets collected for object detection are usually photographed with sufficient ambient lighting. However, their labeling conventions typically focus on clear objects and ignore blurry and occluded objects. Consequently, the detection performance levels of traditional vehicle detection techniques are limited in nighttime environments without sufficient illumination. When objects occupy a small number of pixels and the existence of crucial features is infrequent, traditional convolutional neural networks (CNNs) may suffer from serious information loss due to the fixed number of convolutional operations. This study presents solutions for data collection and the labeling convention of nighttime data to handle various types of situations, including in-vehicle detection. Moreover, the study proposes a specifically optimized system based on the Faster region-based CNN model. The system has a processing speed of 16 frames per second for 500 × 375-pixel images, and it achieved a mean average precision (mAP) of 0.8497 in our validation segment involving urban nighttime and extremely inadequate lighting conditions. The experimental results demonstrated that our proposed methods can achieve high detection performance in various nighttime environments, such as urban nighttime conditions with insufficient illumination, and extremely dark conditions with nearly no lighting. The proposed system outperforms original methods that have an mAP value of approximately 0.2.

In recent years, traffic flow controlling and monitoring have become important issues in intelligent transportation systems (ITSs). For example, vehicle flow management can positively benefit the ITSs from accurate road vehicle detection and counting surveillance systems. In this paper, a high-accuracy vehicle detection mechanism with robust and adaptive vehicle mask training is proposed to effectively detect vehicles in nighttime scenarios. First, a preprocessing step is applied to segment vehicle lamps and filter out reflections. Next, the similarity and symmetric degree of light components are classified into vehicle lamp candidates. A simple fog lamp elimination method is proposed for adapting the proposed method to various road conditions. Then, the mask is trained by a combination of color distribution and weighting information, and it is utilized in the first stage of vehicle detection. The experimental results verify that the proposed method is effective and adaptive for the detection of vehicles at night. [ABSTRACT FROM PUBLISHER]

This paper presents an effective traffic surveillance system for detecting and tracking moving vehicles in nighttime traffic scenes. The proposed method identifies vehicles by detecting and locating vehicle headlights and taillights using image segmentation and pattern analysis techniques. First, a fast bright-object segmentation process based on automatic multilevel histogram thresholding is applied to effectively extract bright objects of interest. This automatic multilevel thresholding approach provides a robust and adaptable detection system that operates well under various nighttime illumination conditions. The extracted bright objects are then processed by a spatial clustering and tracking procedure that locates and analyzes the spatial and temporal features of vehicle light patterns, and identifies and classifies moving cars and motorbikes in traffic scenes. The proposed real-time vision system has also been implemented and evaluated on a TI DM642 DSP-based embedded platform. The system is set up on elevated platforms to perform traffic surveillance on real highways and urban roads. Experimental results demonstrate that the proposed traffic surveillance approach is feasible and effective for vehicle detection and identification in various nighttime environments. [ABSTRACT FROM AUTHOR]