Here we describe two innovative geo applications using the Oak-D Lite stereo camera containing the Intel Movidius Myriad X Vision Processing Unit. Some of this work is described in Kemeny and Kim (2023). In addition, we reference the work of Jaykumaran (2024) who provides a thorough introduction to the Oak-D Lite camera and describes object detection applications with it including highway pothole detection. We also mention the work of Dunne et al. (2023) from Ubotica, a company installing innovative edge-ai vision systems containing the Myriad X chip on smallsats for RGB object detection (ships at sea for instance), hyperspectral ground/ocean classification (floral outbreaks in the ocean for example), and thermal classification (early solar flare detection for example).

The Oak-D Lite is a compact camera contain 3 cameras (2 for depth, 1 for object detection/classification) and the Myriad X vision processer. It is able to provide real-time vision and depth maps, as well as real time neural net inferencing using embedding neural net routines. It is very compact as shown in the figure below.

The second picture shows Yolo v3 object detection installed on the Myriad X chip showing xyz distances to labeled objects.  Calibration tests were conducted to determine object size from pixel area and distance, with errors less than 15% at distances up to 20 feet. We are currently training a Yolo v8 model to identify rockfall on or adjacent to highways, with a drone carrying the Oak-D Lite along with a Raspberry Pi chip.

A second geo application involves pointing the camera at a rock face to determine the orientation of rock discontinuities (from drone, dashcam, by hand). The Oak-D Lite camera can generate realtime 3D point clouds and normals for each point. We are developing a fast algorithm to find patches of normals with similar orientation (indicating a flat rock discontinuity) and determining the dip and dip direction of each patch.

If you are interested in this topic and want to work with us on further developing it (as research or a potential commercial product) please contact and let’s discuss.

Kemeny, J. and K. Kim. 2023. Simple Sensor Solutions (With the Help of AI) for Geologic and Hydrologic Hazards Associated with Climate Change, Poster presentation at the 2023 Fall Meeting of the American Geophysical Union, San Francisco, CA.

Jaykumaran. 2024. Object Detection on Edge Device – Deploying YOLOv8 on OAK-D-Lite, learnopencv.com/object-detection-on-edge-device/

Dunne, J. Romero-Cañas, S. Caulfield, S. Romih and J. L. Espinosa-Aranda, “Intelligent Space Camera for On-Orbit AI-Driven Visual Monitoring Applications,” 2023 European Data Handling & Data Processing Conference (EDHPC), Juan Les Pins, France, 2023, pp. 1-4, doi: 10.23919/EDHPC59100.2023.10396125.