An ONNX-optimized version of the Depth Anything V2 Small model for efficient inference in production environments.