Eizen SDK Testing Scenarios
Scenario 1 OnPrem : SDK + Eizen Hyd GPU
Scenario 2 Cloud: SDK + Cloud (L-Dev)
| Testings: |
- As a user, I need a basic understanding of terms such as tenant, analytics types, analytics categories, and sources in the usage documentation.
- I need to obtain credentials or an access token from the EIZEN team as well as tenant IDs.
- Certain tasks should be documented in the usage guide so we don’t have to disturb the team every time (e.g., how to create a user for a tenant).
- Creating another category of analytics type for a specified tenant is an issue.
- I cannot create a user because there is no documentation — this is also an issue. |
Scenario 1 SDK + Eizen Hyd GPU | Scenario 2 OnPrem : SDK + AWS | ||||
|---|---|---|---|---|---|
| 1 | VIP | VIP Installation in local GPU | |||
| Whenever we go to a client, they provide the GPU | VIP Installation in Cloud | ||||
| (For now, we use L-Dev) | |||||
| 2 | Real-Time Analytics | 2.1 Configure the cameras | |||
| → Scanning the cameras/sources | |||||
| → User Admin & Password for cameras | |||||
| Take our local camera and configure it | |||||
| 2.2 Configure the models | |||||
| → Person Detection | |||||
| 2.3 Store the Analytics | |||||
| MongoDB - Local | |||||
| MongoDB - Cloud | Testing Observations: |
- Source with local camera URL (rtsp) added with person detection model.
- Configured the source with a local MongoDB URL with IP exposed to store the analytics in the GPU and local system.
- Now analytics are stored in both MongoDB's rawanalytics collection.|||
|3|Off-Line Analytics|3.1 Upload the Video (Last 2 months' video)
3.2 Configure the AI Models
3.3 Store the Analytics
MongoDB - Local
MongoDB - Cloud|1. A new Source with video (s3 link) added with model (person detection). - Configured with local MongoDB.
- Analytics are getting stored for the source in both collections in the GPU and locally.|||
|4|Training AI Model|4.1 Create a model by uploading the model data zip file (Yolo-Detections).
4.2 Once training is completed, assign the model to the user. Get the Model Details (Model Id).
4.3 Configure the model to source for Analytics.|1. Trained the model with person detection data. - The model is added to the tenant at the start of the training with an isActive flag set to false.
- The model is updated by Jenkins and deployed (weights are added to the automated deployment folder).
- The model is configured to a source with modelId.
- Activities and events of the model are now configured manually to get analytics.|||
|5|User Admin Creation & User Add|5.1 Adding the User with required privileges.|1. Added the VIP Admin and created a tenant manually. - Added a User using tenant access with a “User” role.
- When added, the user creates a temporary password which is updated manually in Keycloak to a permanent password.|||
|6|Adding Analytics, Zone|6.1 Adding the Analytics type, category, and analytics.
6.2 Add Zones to the analytics|1. Added Analytics type, category, and analytics to a tenant using VIP admin access. - A zone for analytics was added using tenant access.|||
|7|Inferencing the model.|7.1 Infer the model for images or videos directly.|1. Inferred using Person detection model with image and video. - For the image used, input type s3link and output as s3url and detections.
- For video, both input and output are S3 links.|||
|7.1|Model Assigning to the tenants.|7.1.1 Assign the created model to the Tenant user.|1. When the training started, we added the model details to the models table with the isActive flag set to false. - By taking the modelId and tenantId, we are assigning the model to the tenant at the start of the training.|||
|8|Retraining the Model.|8.1 Retraining the model with new Data.|1. Retrained the model using the model ID and new data. - Trained using both variables and yaml file.|||
|9|Data Collection.|9.1 Collecting the data for annotations from the source.|1. Collected the data from the source into Label Studio. - When saving the images from S3 link, they were not properly stored in LabelStudio.|||
|10|Usage Credits.||User-level Credits.|||
| Feature | |
|---|---|
| 1 | Image Chat |
| 2 | AI Market Place |
| 3 | Video Summary - text |
| 4 | Video Chat |
| 5 | Sharing & Downloading Reports |
| 6 | Search in Video |
| 7 | Generate Reports - PPT/ PDF |
| 8 | Text-to-video Generation |
| 9 | Chat with reports |
| 10 | List of generated reports |
| 11 | Chat with reports |
| 12 | Alerts Configuration |
| 13 | Search Results in Video Search |
| 14 | Recording Videos |
| 15 | Video Analytics using Multiple Overlapping Cameras |
| 16 | Multi-Tenant Service |
| 17 | Analytics |
| 18 | Video Library |