Jobs
Deploying models and pipelines in Cordatus is a fast and easy process that takes just a few clicks. Each deployment is managed as an independent Job instance. Inference analytics and alarms are tracked and stored separately for each job.
Cordatus uses NVIDIA DeepStream SDK for Intelligent Video Analytics. The DeepStream version may vary between devices, and this can create small differences in Cordatus's capabilities.
Starting a New Job
The deployment process is exactly the same for ready-to-use models, custom models, and pipelines in Cordatus.
Important Notes
- You can create multiple job instances from one model or pipeline
- A job can perform inference on multiple camera streams within your device's license capacity
NOTE: Before deploying a model, ensure you have added at least one active camera and client device to Cordatus. For detailed instructions, see the Camera and Device connection sections.
Job Startup Steps
- Go to the AI Platform → AI Workflow page from the left sidebar of the Web App.
- If you want to start a job with a single model, you can select a model or go to the startup screen via the play button next to it.
- You'll be greeted with the Select a Device screen. Select which device you want to start the job on through this screen, then click the Select Device button.
- Enter the job requirements:
Requirements
| Field | Description |
|---|---|
| Job Name | Give your job a name. The job will be identified by this name in analytics screens. |
| Camera Selection | Select the camera(s) you want to start your job on. You can visit theCameras page to add cameras. |
| Model Assignment Matrix | This matrix allows you to select which model will run on which camera. This screen will only appear if you select multiple models. If you select a single model, this screen will not appear. |
| Computing Device | If your selected device is ready, you'll see theDevice Ready message on the right. |
| GPU Configuration | When you have multiple GPUs, they will be listed here. You can select which GPU you want to start this job on. |
| Add Alarm (Optional) | You can add Model-Based Alarms when starting your job. These alarms need to be defined beforehand. You can follow theAlarms page for this. |
| Schedule Job (Optional) | If you want the job to run and stop at specific times, you can schedule your job. |
- Then click the Start Job button to start the job.
First Deployment
INFO: When deploying a job for the first time on a device, the process may take time depending on the device and internet speed. During the first deployment, the client will download and install the 35GB inference engine to the device. Users can track the download progress by clicking the download icon in the upper right corner of the Client interface. Once completed, the job will start automatically.
If the model hasn't been downloaded before, Cordatus will download the model, build it, connect cameras, and start the inference process. When a job is first started, the model building and downloading model sections may take some time depending on the device and internet speed. After that, a built and downloaded model will not be built again.
NOTE: A new job can only be started from the Web App. However, you can restart, stop, or delete a job from the Client.
Stream Page
On the stream page, you can observe your inferred video views in real-time. Actions you can perform on this page:
- Metrics Icon (upper right): View device metrics for the inference job
- Detection Results Toggle (Center): Turn the display of inference results on/off
- View Mode (lower right): Switch between picture-in-picture and full-screen mode
- Pause/Play Icon (lower left): Stop or continue the video
- All Detections (bottom bar): View time periods when Cordatus detected objects
Viewing a Job
Stream Details
You can see more details about the stream by clicking the graph icon in the upper right corner of the screen. In the opened panel, you can check:
- Frame rate (FPS)
- Bitrate
- Bandwidth
- Resolution
Stream Details
Stopping a Job
You can stop a job by clicking the Stop button under the Actions column in the Web App. Cordatus will stop the job but will not delete the job and related data. This way, you can restart the job later.
Stopping a Job
You can also stop a job by pressing the stop button from the Client.
Stopping a Job on Client
Restarting a Job
You can start a job in Stopped or Error status. For more information about job status codes, see the Job Status Codes section.
Restarting a Job on Web App
Click the Run button under the Actions column.
Restarting a Job
Restarting a Job on Client
Click the Run button under the Actions column.
Restarting a Job on Client
Deleting a Job
You can delete a job by clicking the Delete button under the Actions column.
Deleting a Job
You can also delete the job on the Client by clicking the Delete button under the Actions column in the job row.
Deleting a Job on Client
WARNING: This operation permanently deletes the job and all associated inference analytics data from the client device. This action cannot be undone once deleted. Deleting the job has no effect on the deployed model or pipeline.
Job Status Codes
Cordatus assigns various statuses to jobs throughout their lifecycles. These status codes provide information about whether the job is running as expected. Status codes are displayed in the first column of the job row labeled "Status".
| Status | Description |
|---|---|
| Downloading Model | Model is being downloaded to the device. Occurs during first deployment. |
| Building | Occurs only during the first startup of the model. Inference engine is building the model or pipeline. |
| Preparing Engine | Pipeline is set up and engine is starting. |
| Warming Up | Engine is ready, pipeline is being prepared and connecting to cameras. |
| Running | Job is running as expected. |
| Stopped | User manually stopped the job. |
| Error | An error occurred during job execution. See the troubleshooting section for possible issues. |
Status Codes
Using Interface
The Jobs page offers user-friendly features to improve the overall experience.
Job Information
On this page, you can access detailed information about your jobs: You can use the edit button in the upper right to open and close columns.
- Status: Current state of the job
- Job Name: Name of the job
- Applications: Model(s) used
- Cameras: Connected cameras
- Device: Device running the job
- Container Name: Inference engine used
- Initialized At: Start time
- Last Updated: Update time
Jobs Page
You can use the same features on the Client (except device name, because you're already viewing that device's jobs).
Jobs Page on Client
Search Feature
You can easily find specific jobs by entering the job name using the search box in the upper right corner of the screen.
The same search box is also available on the Client.
Sorting
You can sort your jobs by relevant column data by clicking column headers. This feature is available on both the Client and Web App.
Child Jobs
When you click on a job in the Web App, a new modal opens at the bottom of the page displaying child jobs associated with the selected job. In this modal:
- You can review the job history (how many times the job was started)
- You can get information about child jobs (their statuses, start and end times)
- You can delete child jobs by clicking the delete icon under the Actions column
Child Jobs on Web App
When you click on a job on the Client, the same modal opens in the center of the screen with the same functionalities.
Child Jobs on Client
Version Support
| Platform | OS / L4T | JetPack | AGX Thor | AGX Orin Industrial | AGX Orin | Orin NX | Orin Nano | Xavier Series | TX2 Series |
|---|---|---|---|---|---|---|---|---|---|
| ARM64 (Jetson) | L4T 38.2 | JetPack 7.0 | ✔️ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ |
| ARM64 (Jetson) | L4T 36.4.4 | JetPack 6.2.1 | ❌ | ✔️ | ✔️ | ✔️ | ✔️ | ❌ | ❌ |
| ARM64 (Jetson) | L4T 36.2 | JetPack 6.0 | ❌ | ✔️ | ✔️ | ✔️ | ✔️ | ❌ | ❌ |
| ARM64 (Jetson) | L4T 35.6.2 | JetPack 5.1.5 | ❌ | ✔️ | ✔️ | ✔️ | ✔️ | ✔️ | ❌ |
| ARM64 (Jetson) | L4T 35.4.1 | JetPack 5.1.2 | ❌ | ✔️ | ✔️ | ✔️ | ✔️ | ✔️ | ❌ |
| ARM64 (Jetson) | L4T 35.3.1 | JetPack 5.1.1 | ❌ | ✔️ | ❌ | ✔️ | ✔️ | ✔️ | ❌ |
| ARM64 (Jetson) | L4T 35.2.1 | JetPack 5.1 | ❌ | ✔️ (32GB) | ❌ | ✔️ (16GB) | ❌ | ✔️ | ❌ |
| ARM64 (Jetson) | L4T 35.1 | JetPack 5.0.2 | ❌ | ✔️ (32GB) | ❌ | ❌ | ❌ | ✔️ | ❌ |
| ARM64 (Jetson) | L4T 32.7.6 | JetPack 4.6.6 | ❌ | ❌ | ❌ | ❌ | ❌ | ✔️ | ✔️ |
| ARM64 (Jetson) | L4T 32.7.1 | JetPack 4.6.1 | ❌ | ❌ | ❌ | ❌ | ❌ | ✔️ | ✔️ |
| x86_64 | Ubuntu 22.04 | — | — | — | — | — | — | — | — |
| x86_64 | Ubuntu 24.04 | — | — | — | — | — | — | — | — |
For detailed information about the DeepStream version installed on your device, you can visit the NVIDIA DeepStream SDK page.
Copyright © 2025 Cordatus.