We are happy to share the excitement from the article which NVIDIA, well-known GPU’s provider and AI focused tech company, published about DataFromSky – our AI in the Sky!
As the article further explains, we train our deep neural networks using thousands of hours of video footage from around the globe, shot in various weather conditions. The training takes place on NVIDIA GPUs using Caffe and TensorFlow. These specialized neural networks can then recognize objects of interest and continually track them in video footage. The data captured via this AI process is used in numerous research projects, enabling deeper analysis of object interaction and new behavioral models of drivers in specific traffic situations. Ultimately, this kind of data will also be crucial for the development of autonomous vehicles!
We are also carrying out extensive optimizations of the whole architecture, moving to NVIDIA Jetson AGX Xavier, in order to be able to analyze data in real time. Our current work focus is also on the development of an online monitoring drone, which will be able to evaluate image data in real-time, directly during its flight. This type of drone can be used, for example, as a substitute for a police helicopter, or can operate as a mobile surveillance system for personal and property protection.
Read the full article on NVIDIA blog at this link.
We are also a member of the NVIDIA Inception program – virtual accelerator program led by NVIDIA nurtures dedicated and exceptional technological companies, who are revolutionizing industries with advances in AI and data science and help them in all stages of product development, prototyping, and deployment.
Our partner Roelofs has made an analysis with DataFromSky for a big junction in the city center of Brussel, Belgium. The analysis covers the junction between the Leopold II Lane and Sainctelette – one of the biggest and most complex junctions in Brussel. In this case, the video is made by a tethered drone to get a longer survey. Thanks to DataFromSky, it is possible to create a detailed traffic study which was handled to the government of Brussel and its Mobility department, providing the correct data for creating solutions that can improve the traffic flow and safety of this junction.
DataFromSky Aerial analysis is providing complete and detailed telemetry data about each traffic participant:
complete object trajectories (time, position, speed, acceleration in coordinate system UTM or WGS84) – visualization in program or various exports
classification of all objects into 7 main categories (including bicycles and pedestrians)
vehicle counts via configurable counting gates
detailed OD matrix and turning movements count in defined directions
speed and acceleration anywhere in the scene, heatmaps
measurement of critical headway (gap) and follow-up headway
safety analysis (Time to collision measurement, PET, Heavy breaking etc.)
and much more!
Do you want to use DataFromSky project in your project? Contact us with your query!
DataFromSky delivers the data for the models of tactical behavior of cyclists communicating their intentions.
The research project focuses on modeling the behavior of bicyclists at intersections under complex traffic situations and interactions with motor vehicle traffic.
Bicycling is an important part of the urban transport system and traffic control measures are necessary to control traffic flow. The capacity and traffic quality of all traffic participants at signalized intersections can be highly influenced depending on the volume of bicycles.
Video data for this project were collected by setting up high definition cameras on the top of the buildings (to obtain suitable aerial view for the DataFromSky Aerial analysis) near several intersections in the cities of Munich and Freiburg in Germany. Explicit and implicit behavior patterns of cyclists at intersections in the video are observed. Explicit behaviors include bicyclists’ body language to reveal their intentions like hand gestures or head movement. Implicit behaviors such as speed change, view change, or path change, indicate the lateral movement of cyclists indirectly. As for lane changing behavior and driving features, such as speed and positions, extracted by DataFromSky, the cyclist’s path selection behavior at intersections is analytically described.
The analyzed video data results in a model that predicts the behavior of cyclists at intersections based on their explicit and implicit behavior patterns in specific traffic situations. The resulting model can be used for example for supporting tactical and operational functions of automated vehicles in specific traffic situations or as an initial analysis that can give traffic and urban planners a very good overview for proposing some traffic adjustments.
This research project, with the use of DataFromSky, is being carried out by Technical University of Munich in Germany, Department of Civil, Geo and Environmental Engineering, lead by Dr. rer. nat. Andreas Keler, Dr.-Ing. Matthias Spangler, created by M.Sc. Georgios Grigoropoulos and Niu Ruixin.
Traffic flow based on trajectries: green – cars, red – cyclists, blue – pedestrians
In many cities, there are places where many road users pass or stay in a complicated network of routes and get into a risky interaction. It can be difficult to get an overview of the whole situation and define the “hot spots” where the traffic behavior is simply non-standard. Our tool Data From Sky is in most cases be able to uncover, sort and analyze all movements of vehicles, cyclists and pedestrians in a larger area and evaluate critical parts of a road network, that might be risky for any traffic participant. This kind of overview and analysis can give traffic and urban planners a very good starting point for designing any adjustments.
The figure below shows an area in the city of Aarhus in Denmark and displays the flow of cyclists (pink trajectories) and pedestrians (green trajectories) in the 2-hour period. The main goal of the analysis is to provide the Aarhus Municipality with an overview of cyclists and pedestrian movement in the analyzed area and identify some patterns in their traffic behavior.
After having formed a detailed overview of all the different flows and routes, author of this study (Birger Villadsen, Traffic Specialist at COWI) has focused on those areas, where cyclists’ and pedestrians’ behavior gets into a collision. The detailed DataFromSky Analysis helped to analyze the problematic spots and draw up proposals for changes, that can help to simplify the traffic flow. Aarhus Municipality will evaluate the list of proposals and test them on the site by making temporary changes and observing whether they have the intended effect before the changes are implemented permanently.
*full article can be found at Cowi Creators blog.