Progress Report: 3D models from user videos

Initially coming from the idea of using drone images for 3D reconstruction of public places and monuments (project descriptionreport 1) we re-defined the goal to creating a web service to perform 3D reconstruction for any type of video uploaded by a user (report 2). In a crowdsourcing manner users would have to share information such as the location where the footage is from (famous point of interest or GPS coordinates). Our intention was thereby to use the collected data for map enhancement purposes. Other projects might build up on our work and either integrate generated 3D models into map applications like openstreetmap or fuse models taken from neighboring locations to create larger 3D models.

Updated progress plan:

We are currently working on milestone #4 Web interface [5 weeks]. Having a complete 3D reconstruction pipeline set up, we tackled front-end development for a web service offering user registration, video upload and notification on completion. Given that there are still two weeks left in the time we allocated for this task, we are confident to complete it in time. Currently, the following modules are implemented:

User management: Users can create their own profile, where they manage videos and personal information.

Video upload module: Users can upload videos including several optional fields (exact location including GPS, short description, title). Coordinates are  choose manually by adding point on Google Maps and assigning  position on it. Exact location is calculated automatically based on the point position on the map.

Below are  screenshots to give a notion of how the interface is going to look like:

Screen Shot 2015-04-22 at 10.23.40 PM Screen Shot 2015-04-22 at 10.29.06 PM Screen Shot 2015-04-22 at 10.33.02 PM

The modules Videos and 3D model listing and details and Notification  after reconstructing 3d model. In the next step, we want to connect the reconstruction pipeline in the back-end to the web interface.

To prepare for milestone #5 Making our service publicly available  [2 weeks], we now need access to a server that can host our web page and that is capable of executing the reconstruction pipeline. A non-negligable amount of time will be needed to install the required libraries. We cannot make assumptions about expected numbers of users or videos per day. On the other hand, we are rather interested in implementing the concept than in optimizing for a particular usage profile. Therefore, the interaction between fron-end and back-end will be kept in a simple serial way. For example, once a video is uploaded the reconstruction starts. Videos arriving while the construction is busy will be treated as soon as there are resources available. This concept can, however, be parallelized depending on the resources we receive to install our service on.

#6 Testing [optional]

Depending on the performance we reach after deploying our idea on the servers we will advertise our service to a smaller or larger number of people through mailing lists and announcement on social media.

Intermediate resume:

We are very happy that our project has developed as planned since the last progress report. As the remaining practical steps only contain a little more effort in development and most of the effort will be spent on administrative tasks (deployment on a server) we do not expect any major risks from now on.