--- library_name: pytorch license: other tags: - android pipeline_tag: depth-estimation --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything_v3/web-assets/model_demo.png) # Depth-Anything-V3: Optimized for Qualcomm Devices Depth Anything 3 (DA3), a model that predicts spatially consistent geometry from arbitrary visual inputs, with or without known camera poses. This is based on the implementation of Depth-Anything-V3 found [here](https://github.com/ByteDance-Seed/Depth-Anything-3). This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/depth_anything_v3) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary). Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device. ## Getting Started There are two ways to deploy this model on your device: ### Option 1: Download Pre-Exported Models Below are pre-exported model assets ready for deployment. | Runtime | Precision | Chipset | SDK Versions | Download | |---|---|---|---|---| | ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything_v3/releases/v0.51.0/depth_anything_v3-onnx-float.zip) | QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything_v3/releases/v0.51.0/depth_anything_v3-qnn_dlc-float.zip) | TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything_v3/releases/v0.51.0/depth_anything_v3-tflite-float.zip) For more device-specific assets and performance metrics, visit **[Depth-Anything-V3 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/depth_anything_v3)**. ### Option 2: Export with Custom Configurations Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/depth_anything_v3) Python library to compile and export the model with your own: - Custom weights (e.g., fine-tuned checkpoints) - Custom input shapes - Target device and runtime configurations This option is ideal if you need to customize the model beyond the default configuration provided here. See our repository for [Depth-Anything-V3 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/depth_anything_v3) for usage instructions. ## Model Details **Model Type:** Model_use_case.depth_estimation **Model Stats:** - Model checkpoint: da3-small - Input resolution: 518x518 - Number of parameters: 24.7M - Model size (float): 94.3 MB ## Performance Summary | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |---|---|---|---|---|---|--- | Depth-Anything-V3 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 33.653 ms | 5 - 645 MB | NPU | Depth-Anything-V3 | ONNX | float | Snapdragon® X2 Elite | 32.403 ms | 71 - 71 MB | NPU | Depth-Anything-V3 | ONNX | float | Snapdragon® X Elite | 77.059 ms | 70 - 70 MB | NPU | Depth-Anything-V3 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 55.141 ms | 0 - 815 MB | NPU | Depth-Anything-V3 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 75.937 ms | 0 - 79 MB | NPU | Depth-Anything-V3 | ONNX | float | Qualcomm® QCS9075 | 97.015 ms | 3 - 9 MB | NPU | Depth-Anything-V3 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 41.438 ms | 2 - 642 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 42.341 ms | 3 - 577 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Snapdragon® X2 Elite | 44.403 ms | 3 - 3 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Snapdragon® X Elite | 101.353 ms | 3 - 3 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 77.829 ms | 3 - 731 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 223.72 ms | 1 - 551 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 102.425 ms | 3 - 6 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Qualcomm® SA8775P | 109.455 ms | 0 - 546 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Qualcomm® QCS9075 | 128.486 ms | 3 - 9 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 170.993 ms | 3 - 693 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Qualcomm® SA7255P | 223.72 ms | 1 - 551 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Qualcomm® SA8295P | 141.266 ms | 1 - 539 MB | NPU | Depth-Anything-V3 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 57.121 ms | 0 - 567 MB | NPU | Depth-Anything-V3 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 30.907 ms | 1 - 486 MB | NPU | Depth-Anything-V3 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 55.095 ms | 1 - 614 MB | NPU | Depth-Anything-V3 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 184.45 ms | 1 - 473 MB | NPU | Depth-Anything-V3 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 76.653 ms | 1 - 12 MB | NPU | Depth-Anything-V3 | TFLITE | float | Qualcomm® SA8775P | 83.506 ms | 1 - 473 MB | NPU | Depth-Anything-V3 | TFLITE | float | Qualcomm® QCS9075 | 96.949 ms | 0 - 80 MB | NPU | Depth-Anything-V3 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 121.152 ms | 2 - 623 MB | NPU | Depth-Anything-V3 | TFLITE | float | Qualcomm® SA7255P | 184.45 ms | 1 - 473 MB | NPU | Depth-Anything-V3 | TFLITE | float | Qualcomm® SA8295P | 105.618 ms | 1 - 472 MB | NPU | Depth-Anything-V3 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 39.651 ms | 1 - 484 MB | NPU ## License * The license for the original implementation of Depth-Anything-V3 can be found [here](https://github.com/ByteDance-Seed/Depth-Anything-3/blob/main/LICENSE). ## References * [Depth Anything 3: Recovering the visual space from any views](https://arxiv.org/abs/2511.10647) * [Source Model Implementation](https://github.com/ByteDance-Seed/Depth-Anything-3) ## Community * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI. * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).