Technology

DEPZ cameras provide AI-powered stereo vision and depth estimation. We calculate depth based on the latest progress in AI, with neural networks trained to work best with our cameras. This produces significantly higher quality depth maps compared to classical methods. Another advantage is easy adaptation to new environments by fine-tuning models with new data.

Our philosophy: a camera should be a camera, a computer should be a computer. Our cameras deliver synchronized stereo images with calibration data—your device handles rectification and AI processing. Use rectification and AI models from our SDK or use your own code.

We support inference on Bare CPU (x64, ARM), GPU (Nvidia CUDA, AMD ROCm), NPUs (Intel NPU, AMD NPU, RKNN) as well as wide variety of devices like Nvidia Jetson, Raspberry Pi, Radxa and Orange Pi SBCs.

Our SDK includes cutting-edge depth estimation models developed in-house. We provide models for different use cases and hardware—CPU, GPU, and NPU. Some models prioritize quality with higher computational cost, while others are optimized for high FPS.

All models are general-purpose and work well both indoors and outdoors. We also fine-tune models for corporate clients upon request.

Our stereo cameras provide synchronized RGB frames from both sensors with calibration info via USB using UVC protocol. Just plug and play! We believe that this is the closest to natural and the most efficient way to estimate depth and do computer vision for many applications.

Each camera is factory-calibrated for accurate depth measurements. Calibration data is stored on our servers and can be accessed through our SDK. For clients without internet access due to policy or use case requirements, calibration data can also be downloaded once and stored locally.

Our SDK includes smart re-calibration flow. It adjusts calibration to keep precise depth during all camera lifetime.

We designed our technology pipeline to provide users with as many camera and module versions as possible. Our cameras offer a choice of different lenses, baselines, and mounting options.

We also provide custom camera development services for specialized requirements.

We strive to make our cameras as friendly as possible for laboratories and R&D departments in universities, startups, and large companies. We provide a wide range of accessories and mounting options, and we are always ready to answer questions or support new hardware platforms and operating systems.

Our SDK provides access to calibration info, raw stereo images, rectified stereo images, disparity maps, depth maps and point clouds.

We also provide custom vision software development services for integration into real-world systems, new platforms, and new use cases.