Depth-Anything-V2: Optimized for Qualcomm Devices
Depth Anything is designed for estimating depth at each point in an image.
This is based on the implementation of Depth-Anything-V2 found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.
Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up 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 |
| QNN_DLC | float | Universal | QAIRT 2.45 | Download |
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | Download |
| TFLITE | float | Universal | QAIRT 2.45 | Download |
For more device-specific assets and performance metrics, visit Depth-Anything-V2 on Qualcomm® AI Hub.
Option 2: Export with Custom Configurations
Use the Qualcomm® AI Hub Models 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-V2 on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.depth_estimation
Model Stats:
- Model checkpoint: DepthAnything_V2_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-V2 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 30.194 ms | 5 - 534 MB | NPU |
| Depth-Anything-V2 | ONNX | float | Snapdragon® X2 Elite | 30.425 ms | 49 - 49 MB | NPU |
| Depth-Anything-V2 | ONNX | float | Snapdragon® X Elite | 73.819 ms | 48 - 48 MB | NPU |
| Depth-Anything-V2 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 48.977 ms | 0 - 667 MB | NPU |
| Depth-Anything-V2 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 72.369 ms | 0 - 58 MB | NPU |
| Depth-Anything-V2 | ONNX | float | Qualcomm® QCS9075 | 90.33 ms | 3 - 9 MB | NPU |
| Depth-Anything-V2 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 35.95 ms | 2 - 503 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 33.333 ms | 3 - 459 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Snapdragon® X2 Elite | 36.349 ms | 3 - 3 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Snapdragon® X Elite | 92.27 ms | 3 - 3 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 66.376 ms | 0 - 622 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 205.467 ms | 2 - 475 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 91.276 ms | 3 - 6 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Qualcomm® SA8775P | 100.423 ms | 1 - 497 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Qualcomm® QCS9075 | 103.761 ms | 3 - 9 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 141.311 ms | 3 - 536 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Qualcomm® SA7255P | 205.467 ms | 2 - 475 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Qualcomm® SA8295P | 132.831 ms | 0 - 432 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 50.833 ms | 0 - 467 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 89.552 ms | 2 - 1900 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 94.877 ms | 2 - 2 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Snapdragon® X Elite | 154.944 ms | 2 - 2 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 127.19 ms | 1 - 2173 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Qualcomm® QCS8275 (Proxy) | 291.636 ms | 2 - 1928 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Qualcomm® QCS8550 (Proxy) | 168.943 ms | 2 - 4 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Qualcomm® SA8775P | 157.376 ms | 2 - 1779 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Qualcomm® QCS9075 | 159.671 ms | 2 - 5 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Qualcomm® SA7255P | 291.636 ms | 2 - 1928 MB | NPU |
| Depth-Anything-V2 | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 103.043 ms | 2 - 1911 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 30.323 ms | 1 - 467 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 56.941 ms | 1 - 614 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 185.083 ms | 1 - 469 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 79.394 ms | 1 - 12 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Qualcomm® SA8775P | 82.713 ms | 1 - 436 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Qualcomm® QCS9075 | 97.364 ms | 0 - 58 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 154.196 ms | 0 - 593 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Qualcomm® SA7255P | 185.083 ms | 1 - 469 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Qualcomm® SA8295P | 106.525 ms | 0 - 429 MB | NPU |
| Depth-Anything-V2 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 41.489 ms | 1 - 440 MB | NPU |
License
- The license for the original implementation of Depth-Anything-V2 can be found here.
References
Community
- Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
- For questions or feedback please reach out to us.
