1. Video Insights project structure

The process of traffic data extraction from the video files, by which the structured and anonymized vehicle trajectories are generated from the video material together with their position in time and space, is called processing. It’s an automated procedure utilizing the GoodVision proprietary artificial intelligence algorithms and takes only several hours for the outputs to be presented in the Video Insights application. But before the videos can be processed, they have to be uploaded into the app first. Read further to explore the way of providing the video files into the application structure.

For uploading videos to GoodVision Video Insights, you’ ll need a dedicated Project. Project is the highest level of the application structure in which various Locations and Data sources can be created for sorting the video files according to your needs. Read how to organize your recordings in the Video Insights project structure here.

Each Project has its own Vault space where the video files are stored before processing. GoodVision Vault is a secure, intermediate input space for uploading videos to the Projects in the Video Insights application. Videos imported to the Vault are securely stored in the cloud platform and can be played inside the application, downloaded again when needed or sent for processing into a chosen Data Source. Get to know more about the GoodVision Vault in this article.

2. Uploading videos to the Video Insights platform

After you have created a Project, navigate to the Vault tab on the left panel. Here’s where you upload all videos necessary for the traffic analysis. Either simply drag and drop the desired video file(s) from your computer to the Vault or click on the “+” and then “Upload video/folder”. You can also create a folder structure in the Vault for better orientation in the video files and upload the video material directly to the prepared folders and subfolders.

Picture 1: uploading video files to the GoodVision Vault

No matter if it is a single file or a thousands of smaller chunks supposed to be connected together, GoodVision handles all for you automatically. In case you need to analyze very large videos or you are having difficulties uploading loads of data to GoodVision platform via your browser, try our desktop application which allows you to upload unlimited amounts of video data to GoodVision Vault safely and without concerns for connection issues. Check the GoodVision Sync feature for more information.

3. Processing the video files

With the video files successfully imported to the Vault, you can start processing them for the traffic analysis. In fact, you don’t have to wait until the files are completely saved in the Vault, the same process applies also for files which are still being uploaded.

Go to the Project and create one or more Locations inside the Project, depending on your needs and project specifications. Location represents a set of cameras sharing a common feature (e.g. a city, a district, an intersection or a roundabout/street). It is also a folder for your Data Sources - the camera views of your recordings where the videos will be placed after the successful processing. Videos in a Data Source shall be from the identical scene with the same viewpoints and video parameters like resolution or FPS.

Create a Data Source in a selected Location by clicking the “New data source” tab on the right panel. The following window will appear where you’ll fill in the Data Source name, The Vault as a source type and choose the appropriate device type - a fixed camera, a drone camera or a camera with a time lapse footage. You can also add an exact GPS or map location of the Data Source. Read more about GoodVision processing options here.

Picture 2: Adding a new Data Source

When the Data Source is created for your video footage, you can anytime follow with sending the video for processing to the Data Source. Even if the desired video is still uploading, you can click on the green “Process” button and select the video you want to process from the Vault.

Picture 3: An empty Data Source with active “Process” button

A short dialog box will open where you can review the video data like filename, duration, device type and processing cost. What you can change is the video start. Please take a look at the video start time and date as the input there must not be the exact time of the recording but is automatically taken from the video metadata.

You can also decide to delete the original video immediately after the processing. If you choose so, the video will be processed normally but will not be available anymore in the application physically, only the extracted data will. That might save some Vault space on one hand but also has some disadvantages on the other.

When the video is kept in the application, you can review the original footage during analysis directly on the Describe page of the Data Source to check what was really happening on the scene in the given time. This playback feature was designed to bring value especially to traffic modellers who need to quickly jump to a video during the analysis to check the real situation. You can also download the video any time later or share it to your colleagues as part of the project.

Picture 4: Video processing dialog window

Videos sent for processing without deleting them will not be available from the Vault space either but will be moved to the Data Source detail from where they can be downloaded again or deleted from the Vault any time later when they are not needed anymore.

Picture 5: Data source detail with the original video file available after processing

4. How to upload multiple files at once

It is also possible to select multiple files (up to 500 at once) and upload them to a single Data Source. Imagine your cameras are producing short 15 minutes chunks of video, GoodVision will make sure that object trajectories between those videos will be connected without losing accuracy.

Uploading multiple files works in the same way as the single file upload. First, they have to be uploaded to the Vault (preferably to a separate folder for better orientation) and then chosen for processing in a Data Source. Make sure your videos are from the same camera and are directly linked to each other — e.g. there are no time gaps between them, because GoodVision will concatenate the data as if there is no gap. Multi-file upload dialog will allow you to change the order of the files by dragging them but it's easier to have them prepared in a consecutive way.

Picture 6: Processing of multiple consecutive video files

5. Uploading video files to previous uploads

If you need to upload more video files to already existing videos in a selected camera, just proceed the same way as by a regular video upload. The only thing you have to do in this case is to manually set the time of the first new video to upload to the end of the already uploaded video file from before. The system will automatically stitch the following video files to the current upload.

6. Uploading large volumes of video data via GoodVision Sync

If you are in the situation where you want to upload loads of data or very large videos, the GoodVision Sync is a perfect solution for you. The feature is available as part of the Desktop Application.

You just need to prepare a list of multiple video files or respective folders with videos and start uploading them to GoodVision Vault. From that moment you do not need to worry about the process anymore. No matter if ten, one hundred or one thousand of videos, the application will keep uploading them successively on the background whenever the internet connection is active.

7. Video formats supported

Video Insights supports video uploads from various sources, various containers (*.mp4, *.avi, *mov, *.asf .._) and codecs (h.264, h.265, .._). Supported are files with various lengths and sizes up to 100 GB per file.

If your video format is not supported, Video Insights will notify you, won’t process your video and return your credits. To make sure your video format is supported, contact support@goodvisionlive.com.

Read about another way of video input: Time-lapsed video processing

Did this answer your question?