UAV based traffic monitoring holds distinct advantages over traditional traffic sensors, such as loop detectors, as UAVs have higher mobility, wider field of view, and less impact on the observed traffic. For traffic monitoring from UAV images, the essential but challenging task is vehicle detection. This paper extends the framework of Faster R-CNN for car detection from low-altitude UAV imagery captured over signalized intersections. Experimental results show that Faster R-CNN can achieve promising car detection results compared with other methods. Our tests further demonstrate that Faster R-CNN is robust to illumination changes and cars’ in-plane rotation. Besides, the detection speed of Faster R-CNN is insensitive to the detection load, that is, the number of detected cars in a frame; therefore, the detection speed is almost constant for each frame. In addition, our tests show that Faster R-CNN holds great potential for parking lot car detection. This paper tries to guide the readers to choose the best vehicle detection framework according to their applications. Future research will be focusing on expanding the current framework to detect other transportation modes such as buses, trucks, motorcycles, and bicycles.

A new hybrid vehicle detection scheme which integrates the Viola-Jones (V-J) and linear SVM classifier with HOG feature (HOG + SVM) methods is proposed for vehicle detection from low-altitude unmanned aerial vehicle (UAV) images. As both V-J and HOG + SVM are sensitive to on-road vehicles’ in-plane rotation, the proposed scheme first adopts a roadway orientation adjustment method, which rotates each UAV image to align the roads with the horizontal direction so the original V-J or HOG + SVM method can be directly applied to achieve fast detection and high accuracy. To address the issue of descending detection speed for V-J and HOG + SVM, the proposed scheme further develops an adaptive switching strategy which sophistically integrates V-J and HOG + SVM methods based on their different descending trends of detection speed to improve detection efficiency. A comprehensive evaluation shows that the switching strategy, combined with the road orientation adjustment method, can significantly improve the efficiency and effectiveness of the vehicle detection from UAV images. The results also show that the proposed vehicle detection method is competitive compared with other existing vehicle detection methods. Furthermore, since the proposed vehicle detection method can be performed on videos captured from moving UAV platforms without the need of image registration or additional road database, it has great potentials of field applications. Future research will be focusing on expanding the current method for detecting other transportation modes such as buses, trucks, motors, bicycles, and pedestrians.

Conflict analysis using surrogate safety measures (SSMs) has become an efficient approach to investigate safety issues. The state-of-the-art studies largely resort to video images taken from high buildings. However, it suffers from heavy labor work, high cost of maintenance, and even security restrictions. Data collection and processing remains a common challenge to traffic conflict analysis. Unmanned Aerial Systems (UASs) or Unmanned Aerial Vehicles (UAVs), known for easy maneuvering, outstanding flexibility, and low costs, are considered to be a novel aerial sensor. By taking full advantage of the bird’s eye view offered by UAV, this study, as a pioneer work, applied UAV videos for surrogate safety analysis of pedestrian-vehicle conflicts at one urban intersection in Beijing, China. Aerial video sequences for a period of one hour were analyzed. The detection and tracking systems for vehicle and pedestrian trajectory data extraction were developed, respectively. Two SSMs, that is, Postencroachment Time (PET) and Relative Time to Collision (RTTC), were employed to represent how spatially and temporally close the pedestrian-vehicle conflict is to a collision. The results of analysis showed a high exposure of pedestrians to traffic conflict both inside and outside the crosswalk and relatively risking behavior of right-turn vehicles around the corner. The findings demonstrate that UAV can support intersection safety analysis in an accurate and cost-effective way.

Object detection on railway tracks, which is crucial for train operational safety, face numerous challenges such as multiple types of objects and the complexity of train running environment. In this study, a multi-sensor framework is proposed to fuse camera and LiDAR data for the detection of objects on railway track including small obstacles and forward trains. The framework involves a two-stage process: region of interest extraction and object detection. In the first stage, a multi-scale prediction network is designed to achieve pixel level segmentation of the railway track and forward train via the image. In the second stage, LiDAR data is used to estimate the distance to the train and detect small obstacles in the railway track area which is extracted from the first stage. Experimental results show that the region of interest extraction method achieves desirable accuracy for railway track and train segmentation; and the proposed fusion method outperforms the one based on camera or LiDAR alone for small obstacles and forward train detection. Moreover, in practice the proposed framework has been successfully applied on the Hong Kong Metro TSUEN WAN line and the Beijing Metro YANFANG line.

This research develops an advanced vehicle detection method, which improves the original Viola-Jones (V-J) object detection scheme for better vehicle detections from lowaltitude unmanned aerial vehicle (UAV) imagery. The original V-J method is sensitive to objects' in-plane rotation, and therefore has difficulties in detecting vehicles with unknown orientations in UAV images. To address this issue, this research proposes a road orientation adjustment method, which rotates each UAV image once so that the roads and on-road vehicles on rotated images will be aligned with the horizontal direction and the V-J vehicle detector. Then, the original V-J can be directly applied to achieve better efficiency and accuracy. The enhanced V-J method is further applied for vehicle tracking. Testing results show that both vehicle detection and tracking methods are competitive compared with other existing methods. Future research will focus on expanding the current methods to detect other transport modes, such as buses, trucks, motorcycles, bicycles, and pedestrians.

Driven by the prominent thermal signature of humans and following the growing availability of unmanned aerial vehicles (UAVs), more and more research efforts have been focusing on the detection and tracking of pedestrians using thermal infrared images recorded from UAVs. However, pedestrian detection and tracking from the thermal images obtained from UAVs pose many challenges due to the low-resolution of imagery, platform motion, image instability and the relatively small size of the objects. This research tackles these challenges by proposing a pedestrian detection and tracking system. A two-stage blob-based approach is first developed for pedestrian detection. This approach first extracts pedestrian blobs using the regional gradient feature and geometric constraints filtering and then classifies the detected blobs by using a linear Support Vector Machine (SVM) with a hybrid descriptor, which sophisticatedly combines Histogram of Oriented Gradient (HOG) and Discrete Cosine Transform (DCT) features in order to achieve accurate detection. This research further proposes an approach for pedestrian tracking. This approach employs the feature tracker with the update of detected pedestrian location to track pedestrian objects from the registered videos and extracts the motion trajectory data. The proposed detection and tracking approaches have been evaluated by multiple different datasets, and the results illustrate the effectiveness of the proposed methods. This research is expected to significantly benefit many transportation applications, such as the multimodal traffic performance measure, pedestrian behavior study and pedestrian-vehicle crash analysis. Future work will focus on using fused thermal and visual images to further improve the detection efficiency and effectiveness.