How Can RidgeRun Help?

Your edge AI application needs to work with your cameras, models and target hardware. It also needs to meet your frame rate, latency and memory requirements. RidgeRun works with your team to develop the application, connect the devices, and create or optimize the operators you need.

Contact RidgeRun

Contact RidgeRun to see how we can help with your project. You can reach out while choosing an approach, developing an application or improving an existing one.

From a First Pipeline to a Product

A running example is a useful start. Building your application brings questions such as:

How do I connect my camera or sensor to the processing pipeline?
How do I prepare the data for my AI model?
What needs to change to meet my latency, frame rate or memory goals?
How do I connect Holoscan to software I already use?
How do I build, deploy and update the application on the target?

We can work on one of these parts or help develop the complete application.

Where Engineering Support Fits

Sensor and Camera Integration

We integrate cameras and sensors with Holoscan and adapt capture operators to the format and timing your application needs. The work can include drivers, capture configuration, timestamps, and passing frames to processing or inference. We can also connect the pipeline to capture software you already use.

The ArgusCameraOp and V4L2UserPtrSourceOp pages show two RidgeRun capture operators. See Holoscan Camera Overview for the Camera module's capture options.

Operators and Pipeline Architecture

We design application pipelines, develop custom operators and adapt existing processing code to Holoscan. This can include camera preprocessing, model input preparation, display, and connections to C++, CUDA or GStreamer components.

For example, RawBayerToYoloTensorOp combines several camera preprocessing steps into one CUDA operator.

When an application needs separate processes, we can define how they exchange data and how the application handles source changes.

Performance and Memory Paths

We profile the application on the target hardware to find where it spends time and uses memory. We can then work on unnecessary copies, buffer sizes, CUDA processing and the way stages run together. The goal is to help your application meet its latency, frame rate and memory requirements.

The Performance and Debugging guide introduces the checks used to investigate application problems.

Build and Deployment

We can help prepare the build and setup for your target, package the application, and define how it starts, stops and reports problems. We also help with container setup, logging, updates and the tests needed before deployment.

Begin with Build, Run, and Validate for the local build and first-run path.

Migration and Technical Reviews

We review existing applications and help move them to Holoscan 5. The work can include porting operators, adapting connections and timing rules, and checking that the updated application keeps the behavior you need. A review can cover one operator or the complete pipeline.

The Migration Concept Map introduces the application changes to consider.

Examples of RidgeRun Work

The guide includes implementations and measurements you can explore:

Example	What it shows
RidgeRun operators	Camera capture, CUDA preprocessing, inference, display and application utilities.
Interprocess example	Frame exchange between processes and switching between input streams while the receiver keeps running.
Holoscan vs GStreamer benchmark	Latency, CPU, GPU and memory measurements for the tested capture, display and inference pipelines.

What to Bring

Tell us what you want the application to do and where you need help. If you already know the target hardware, cameras, model or performance goals, share those details too. You can also share code or measurements if you have them. You do not need to have all of this ready before contacting us.

Continue Learning

Holoscan SDK at a Glance for the overall learning path.
Build, Run, and Validate for installation and first execution.
Holoscan Camera Overview for camera integration choices.
Vision Pipeline Example for a complete application path.