Car Detection from Low-Altitude UAV Imagery with the Faster R-CNN

Abstract

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.

Informed Region Selection for Efficient UAV-based Object Detectors: Altitude-aware Vehicle Detection with CyCAR Dataset

“…Nevertheless, in vehicle detection from aerial imagery, high accuracy detectors supporting small spatial resolution of objects are essential to permit high-altitude flights, resulting to larger ground-area coverage [8]. Recent research has shown that region-based detectors demonstrate remarkable accuracy under such challenging conditions [25], at the expense of increased computational payload. In the work of [9], the speed-accuracy trade-off in convolutional detectors is broadly discussed.…”

“…In recent literature, this post-processing step has been enhanced to utilise additional application-specific information, in order to eliminate false-positive detections that downgrade the model's accuracy, such as bounding boxes with unexpected size and/or aspect ratio [25] [16]. However, since these methods are only applied on the final output of the predictor, they provide no improvement on its execution time.…”

Classification of Metro Facilities with Deep Neural Networks

“…Reference [10] proposed a DCNN for the detection of arcs in pantograph-catenary systems. Reference [11] used a DCNN for car detection. Although the DCNN has shown impressive results, limited data and high computational resources are barriers to its use [2].…”

Object Detection Based on Fast/Faster RCNN Employing Fully Convolutional Architectures

“…At the same time, the feature classifier is usually a linear SVM [5], a nonlinear boosted classifier [6], or an additive kernel SVM [7]. However, the deep ConvNets have dominated the feature extractor of the modern object detectors in various application scenarios [8][9][10][11]. Aside from being capable of representing higherlevel semantics, ConvNets are also more robust to variance in scale and thus facilitate recognition from features computed on a single input scale.…”