Real-Time Vehicle Detection in Aerial Images Using Skip-Connected Convolution Network with Region Proposal Networks

Maiti, Somsukla; Gidde, Prashant Sadashiv; Saurav, Sumeet; Singh, Sanjay; Sangwan, Dhiraj; Chaudhury, Santanu

doi:10.1007/978-3-030-34869-4_22

Cited by 5 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A slight shakiness in the video footage can lead to large errors in vehicles' trajectories, especially when the videos are taken from an oblique angle and long distance (Barmpounakis et al 2016;Khan et al 2017a). Moreover, detection of vehicles in aerial videos is still an active research problem in computer vision, mainly due to their small D r a f t size with regard to the entire frame and potential interference of vehicles close to each other laterally or longitudinally (Maiti et al 2019). Yet, considerable research has been recently conducted using a variety of methodological approaches and frameworks for the collection and extraction of vehicle trajectory data from UAV videos.…”

Section: Uav Use In Traffic Data Collectionmentioning

confidence: 99%

Using single and multiple unmanned aerial vehicles for microscopic driver behaviour data collection at freeway interchange ramps

Alamry

Hassan

2022

Can. J. Civ. Eng.

View full text Add to dashboard Cite

This paper presents a detailed methodological framework for collecting microscopic driver and vehicle behaviour data over a long road segment with an application to the entire stretch of a freeway ramp segment using single and multiple Unmanned Aerial Vehicles (UAVs). The methodology allows users to collect reliable and complete trajectories of traffic movements at areas with challenging physical characteristics (long road segment, horizontal curvature, changing elevation, and presence of shadow), challenging traffic characteristics (high traffic volume, high speeds, and high speed changes), and restrictive regulations (UAVs prohibited from hovering over the freeway or the right-of-way). Different UAV setups were recommended and can be used depending on the site conditions. Specific commercial software and procedures used to complete the data collection are explained. The methodology was applied at two ramps and verified with speed data acquired from differential GPS receivers using three different error metrics. Results showed good performance of the proposed methodology, including when aerial videos must be taken from oblique angles.

show abstract

Section: Uav Use In Traffic Data Collectionmentioning

confidence: 99%

Using single and multiple unmanned aerial vehicles for microscopic driver behaviour data collection at freeway interchange ramps

Alamry

Hassan

2022

Can. J. Civ. Eng.

View full text Add to dashboard Cite

show abstract

“…Regarding the UAV ecosystem, deep learning is a mainstream tool for numerous applications, i.e., automatic navigation of UAV [19,26,27], object [27] or vehicle tracking [28][29][30], and (moving) object detection [31,32] under real-time constraints [33]. Some works combine both object detection and tracking [34][35][36][37] or implement the automation of aerial reconnaissance tasks [38].…”

Section: Introductionmentioning

confidence: 99%

Inferring Visual Biases in UAV Videos from Eye Movements

Perrin

Zhang

Meur

2020

Drones

View full text Add to dashboard Cite

Unmanned Aerial Vehicle (UAV) imagery is gaining a lot of momentum lately. Indeed, gathered information from a bird-point-of-view is particularly relevant for numerous applications, from agriculture to surveillance services. We herewith study visual saliency to verify whether there are tangible differences between this imagery and more conventional contents. We first describe typical and UAV contents based on their human saliency maps in a high-dimensional space, encompassing saliency map statistics, distribution characteristics, and other specifically designed features. Thanks to a large amount of eye tracking data collected on UAV, we stress the differences between typical and UAV videos, but more importantly within UAV sequences. We then designed a process to extract new visual attention biases in the UAV imagery, leading to the definition of a new dictionary of visual biases. We then conduct a benchmark on two different datasets, whose results confirm that the 20 defined biases are relevant as a low-complexity saliency prediction system.

show abstract

“…A slight shakiness in the video footage can lead to large errors in vehicles' trajectories (Khan et al, 2017b). Moreover, detection of vehicles in aerial video sequences is still an active research problem in computer vision, mainly due to their small size with regard to the entire frame (Maiti et al, 2019). As indicated by Khan et al (2017b), these limitations are the major impediments to making UAV technologies more effective.…”

Section: Problem Definitionmentioning

confidence: 99%

Examining Driver Behaviour at Freeway Ramp Terminals Based on Trajectory Data Collected Using Unmanned Aerial Vehicles and Video Image Processing

Alamry¹

View full text Add to dashboard Cite

This research utilizes Unmanned Aerial Vehicles (UAVs) to examine real-life speeds and behaviour of drivers at freeway entrance and exit ramp terminals. To achieve this goal, a total of 1,127 minutes of high-resolution aerial video data were collected for traffic movements at thirteen single-lane ramp terminals (seven entrances and six exits) using single and multiple UAVs. A complete space-time trajectory was extracted for each vehicle as it moved on the freeway right lane (FRL) or speed-change lane (SCL) and ramp using a combination of computer vision and deep learning tools. The trajectories were processed to extract relevant driver-vehicle behaviour measures (e.g., merging/diverging location, merging/diverging speed, acceleration/deceleration distances, SCL utilization rates, and accepted merging gaps). A descriptive analysis was performed for better understanding of driver behaviour over the entire stretch of the freeway ramp terminal segment, including FRL, SCL, and ramp. The trends of driver behaviour measures and their relationships with the SCL and ramp geometric characteristics were investigated under free-flow and platoon conditions. Results of the descriptive analysis highlighted differences between taper and parallel SCLs in terms of merging/diverging location, merging/diverging speed, and SCL utilization. Observations of data also confirmed the importance of accounting for the effects of ramp controlling features on the behaviour and vehicle acceleration needs, especially at exit ramps.Several statistical models were developed using regression analysis to model drivers' behaviour measures on SCLs and ramps. Moreover, a set of the observed merging accepted gap data were fitted to the models proposed in the literature to check which models provide the best fit. Results revealed that the models developed using trip data from SHRP-2 Naturalistic Driving Study (NDS) database relatively fitted the data better than other models in the literature. This finding is significant in validating the transferability of models developed using the NDS to other study areas iii in North America. Finally, the research concluded with a demonstration of the practical application of the developed regression models in reliability analysis, considering actual drivers' behaviour and speeds on SCLs and ramps.Professor Yasser Hassan for his advice, support, and guidance during my research and study at Carleton University. His priceless advice guided me at various stages of my academic life. Without his continued assistance, support, and encouragement, this work would not have been possible.Thirdly, I would like to express my sincere gratitude to Milton Carrasco, co-founder and CEO of Transoft Solutions, for his unstinting help and support and for providing access to Transoft Solutions platform to process the UAV videos. I would also like to extend my sincere gratitude to Transoft Solutions team for their help in processing the UAV videos. Finally, I would like to acknowledge the financial support given for this thesis by Taibah...

show abstract

Real-Time Vehicle Detection in Aerial Images Using Skip-Connected Convolution Network with Region Proposal Networks

Cited by 5 publications

References 15 publications

Using single and multiple unmanned aerial vehicles for microscopic driver behaviour data collection at freeway interchange ramps

Using single and multiple unmanned aerial vehicles for microscopic driver behaviour data collection at freeway interchange ramps

Inferring Visual Biases in UAV Videos from Eye Movements

Examining Driver Behaviour at Freeway Ramp Terminals Based on Trajectory Data Collected Using Unmanned Aerial Vehicles and Video Image Processing

Contact Info

Product

Resources

About