The students of TU Berlin just completed three AMOS projects in the winter semester. The project summaries below detail the final result for each of the projects:
- Traffic data visualization and extraction (an app to display traffic data effectively),
- Computer vision for sights (an app to present images and information of sights ), and
- BPF superpower (an extension of kinvolk’s Inspector Gadget and Headlamp)
Traffic data visualization and extraction
| Project name | Traffic data visualization and extraction |
| Project mission | The mission of this project is to deliver a proof of concept which displays traffic data in a dynamic, performant, and user friendly way. To achieve this we normalize the provided data from TomTom and Here APIs, store them in a database, and make them accessible via a REST endpoint to a frontend for display in a dashboard. |
| Industry partner | IAV GmbH |
| Team logo |  |
| Project summary | The AMOS Traffic Tracker is an interactive dashboard built to compare traffic data of the two providers TomTom and HERE. Both providers offer information about current traffic incidents. The UI has four parts. The main page consists of an OpenStreetMap to display our data as color coded polylines – the traffic incidents. The data can be selected by cities and filtered by data provider and incident type. The historization view displays the absolute number of incidents over time in a bar chart for a selected city. This view allows the user to see how the data develops over time. Specific timestamps (from the past) can be selected, which display the historized traffic situation in our map. The backend retrieves data from the providers’ APIs for preconfigured cities. The cities can be added or deleted via a REST endpoint for the client in a configuration view. After that, data is normalized to a common format and stored in a relational database. Further, we evaluate the providers’ incident data for commonalities and process statistical data. The data can be received from several REST endpoints. |
| Project illustration |    |
| Team photo |  |
| Project repository | https://github.com/basselfa/AMOS_Team1 |
SightScan: Computer vision for sights
| Project name | SightScan: Computer vision for sights |
| Project mission | Our mission is to create a reliable and robust application in the form of an MVP for IAV. The main features of this product will be a dynamic and adaptable pipeline that gathers attraction data for different cities and a user-side application that takes location and images as inputs to provide them with fast information about what is currently in front of them. |
| Industry partner | IAV GmbH |
| Team logo |  |
| Project summary | SightScan provides an advanced microservice architecture whose main feature is a fully automated pipeline to train and deliver models that detect city landmarks, all from scratch. This pipeline contains each step from image collection to the final trained model, ready for sight detection. Various clients, such as desktop applications or car window overlays, can use these models. Additionally, our team implemented a user interface to interact with the aforementioned automated pipeline. Our application contains options to detect sights inside static images or live video feeds, submit new data to the infrastructure for improving city models. Suggesting new cities for sight detection is also possible. |
| Project illustration |    |
| Team photo |  |
| Project repository | https://github.com/fabian-0520/amos-pj-ws20-21-computer-vision-for-sights |
BPF superpower
| Project name | BPF superpower |
| Project mission | To develop software that will help the users to debug kubernetes clusters and containers without the source code of each software component that is used in deployment. |
| Industry partner | Kinvolk GmbH |
| Team logo |  |
| Project summary | We have focused our efforts on extending two different applications. The first application that we have extended was inspector gadget. We have adapted and added a tool called tracee, developed by aquasecurity, to the suite provided to the user. This also included extending libbpf.go, to include new functions for interoperation with eBPF system and then adding sufficient functionale to tracee in order to be generically usable among already existing library of gadgets. The second application that we have extended was Headlamp. We have implemented an ability to use the Network Policy Advisor from Inspector Gadget from Headlamp, which is a WebUI application, instead of CLI, enabling both remote and simpler usage of the NPA. |
| Project illustration |   |
| Project repository | https://github.com/adriannajuchacz/amos3-headlamp https://github.com/ogozman/amos3-gadget https://github.com/ogozman/amos3-tracee |
Leave a Reply