Processing Images from Emergent Vision Cameras

After configuring the hardware in Graph View’s System subtab, use Graph View's Data Flow subtab to configure a data pipeline that processes images from Emergent Vision cameras. 

Graph View lets you build the data pipeline with eCapture Pro’s built-in plugins that can leverage the power of GPUs to efficiently process camera images in different ways. Graph View provides the following built-in plugins:

• Image Disk Saving: Saves raw image frames to disk for later playback or analysis.
• NVENC Compression: Compresses video using NVIDIA’s H.264/H.265 hardware encoder for efficient storage or streaming.
• Image Display: Displays processed image data in real time on the Live View tab.

You can also use custom image processing plugins that you create with eSDK Pro. eCapture Pro includes a sample custom plugin called libcuda-kernel-plugin.

For an introduction to Graph View’s Data Flow subtab, see Data Flow Graph.

Before you start

Before you can create data pipelines that process camera images, complete the first two Recording Workflow Procedures in eCapture Pro:

• Procedure 1: Add or Load a Project in Project View
• Procedure 2: Configure the System Hardware in Graph View

Data flow troubleshooting

If a pipeline node has alerts, the pipeline will not run until you resolve the issues or disable the node. To disable a node, select it in the Graph panel and use the toggle switch in the Selected Nodes pane.

Saving Images from Cameras to Disks

You can record camera images by saving the raw, uncompressed image data directly to disk. This is the simplest type of data processing pipeline in eCapture Pro. It starts with a camera node and ends with an Image Disk Saving node.

Figure 1: The image saving pipeline

1. In Graph View, select the Data Flow subtab.
2. In the Data Flow Actions panel (left side), click the Add button and select Image Disk Saving under your server. A new Image Disk Saving node appears in the Graph panel (right side).
3. In the Graph panel, select the new node. The Data Flow Actions panel displays its properties in the Selected Nodes pane.
4. In the Selected Nodes pane:
   
   • Select a Storage Device for the Image Disk Saving task.
     Note: The Storage Devices listed are those you added earlier in Procedure 2 on the System subtab.
   • (Optional) Select Direct IO for better write performance.
   • (Optional) Select Multi-Storage to record to multiple disks in parallel for more throughput and capacity.
5. In the Graph panel, connect the camera node(s) to the Image Disk Saving node to complete the data flow pipeline.

Figure 2: Use the Data Flow Actions panel on Graph View’s Data Flow subtab to configure an Image Disk Saving node

Compressing and Saving Images

Use this procedure to compress image streams with NVIDIA’s H.264 or H.265 encoder. The NVENC Compression node helps reduce storage needs and makes it easier to record or stream high-speed camera data. NVENC compression works by sending the camera images to the NVIDIA GPU for encoding.

Figure 3: The image compression and saving pipeline

1. In Graph View, select the Data Flow subtab.
2. In the Data Flow Actions panel, click the Add button and select the NVENC Compression task under your Server. An NVENC Compression node appears in the Graph panel.
3. In the Graph panel, select the new node. The Data Flow Actions panel displays its properties in the Selected Nodes pane.
4. In the Selected Nodes pane, configure the following NVENC Compression parameters, ensuring the values are compatible with the camera parameters configured on Live View:
   
   • Host GPU
   • Width and Height
   • Pixel Format
   • Codec: H264 (AVC) or H265 (HEVC)
   • Bitrate
   • Output Frame Rate
   • Output Type (Disk or RTMP)  and, depending on the selected type, the Storage Device or RTMP URL
5. In the Graph panel, connect the camera node(s) to the NVENC Compression node to establish the data flow.

Figure 4: Use the Data Flow Actions panel on Graph View’s Data Flow subtab to configure an NVENC Compression node

Displaying Processed Images

Use this procedure to display images that have been processed by a custom plugin. The Image Display node lets you view the results of real-time processing—such as filtering or enhancement—in the Live View tab, making it easier to verify and validate the output of your data flow pipeline.

The following figure shows a processed image display pipeline that includes a node from the libcuda-kernel-plugin, a sample custom plugin included with eCapture Pro.

Figure 5: A processed image display pipeline

1. In Graph View, select the Data Flow subtab.
2. Add a node for a custom plugin:
   
   • In the Data Flow Actions panel, click the Add button and select a custom plugin under Server > Plugins. A node for the custom plugin appears in the Graph panel.
   • In the Graph panel, select the new node for the custom plugin. The Data Flow Actions panel displays its properties in the Selected Nodes pane.
   • In the Selected Nodes pane, configure the parameters for the custom plugin’s node, ensuring the values are compatible with the camera parameters configured on Live View.
3. Add a node for the Image Display task:
   
   • In the Data Flow Actions panel, click the Add button and select the Image Display task under your Server. An Image Display node appears in the Graph panel.
   • In the Graph panel, select the new node. The Data Flow Actions panel displays its properties in the Selected Nodes pane.
   • In the Selected Nodes pane, configure the following Image Display parameters, ensuring the values are compatible with the camera parameters configured on Live View:
     
     • Display Resolution
     • Frame Rate
4. In the Graph panel, establish the data flow by connecting the camera node(s) to the custom plugin’s node. Then connect the custom plugin’s node to the Image Disk Saving node.

Figure 6: Use the Data Flow Actions panel on Graph View’s Data Flow subtab to configure an Image Display node

Figure 7: Use the Data Flow Actions panel on Graph View’s Data Flow subtab to configure a custom processing node