Hugging Face's logo

qualcomm
/
Depth-Anything

Depth Estimation
PyTorch
android
Model card Files
xet
 Community
qaihm-bot commited on
Commit
4567a99
·
verified ·
1 Parent(s): 46cd5aa

v0.48.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.48.0 for changelog.

Files changed (1) hide show
  1. README.md +47 -47
README.md CHANGED
@@ -14,7 +14,7 @@ pipeline_tag: depth-estimation
14
  Depth Anything is designed for estimating depth at each point in an image.
15
 
16
  This is based on the implementation of Depth-Anything found [here](https://github.com/huggingface/transformers/tree/main/src/transformers/models/depth_anything).
17
- This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/depth_anything) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
18
 
19
  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.
20
 
@@ -27,25 +27,25 @@ Below are pre-exported model assets ready for deployment.
27
 
28
  | Runtime | Precision | Chipset | SDK Versions | Download |
29
  |---|---|---|---|---|
30
- | ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.47.0/depth_anything-onnx-float.zip)
31
- | ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.47.0/depth_anything-onnx-w8a16.zip)
32
- | QNN_DLC | float | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.47.0/depth_anything-qnn_dlc-float.zip)
33
- | QNN_DLC | w8a16 | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.47.0/depth_anything-qnn_dlc-w8a16.zip)
34
- | TFLITE | float | Universal | QAIRT 2.43, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.47.0/depth_anything-tflite-float.zip)
35
 
36
  For more device-specific assets and performance metrics, visit **[Depth-Anything on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/depth_anything)**.
37
 
38
 
39
  ### Option 2: Export with Custom Configurations
40
 
41
- Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/depth_anything) Python library to compile and export the model with your own:
42
  - Custom weights (e.g., fine-tuned checkpoints)
43
  - Custom input shapes
44
  - Target device and runtime configurations
45
 
46
  This option is ideal if you need to customize the model beyond the default configuration provided here.
47
 
48
- See our repository for [Depth-Anything on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/depth_anything) for usage instructions.
49
 
50
  ## Model Details
51
 
@@ -60,45 +60,45 @@ See our repository for [Depth-Anything on GitHub](https://github.com/quic/ai-hub
60
  ## Performance Summary
61
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
62
  |---|---|---|---|---|---|---
63
- | Depth-Anything | ONNX | float | Snapdragon® X Elite | 79.343 ms | 50 - 50 MB | NPU
64
- | Depth-Anything | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 56.067 ms | 0 - 867 MB | NPU
65
- | Depth-Anything | ONNX | float | Qualcomm® QCS8550 (Proxy) | 75.966 ms | 0 - 57 MB | NPU
66
- | Depth-Anything | ONNX | float | Qualcomm® QCS9075 | 97.967 ms | 3 - 9 MB | NPU
67
- | Depth-Anything | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 41.282 ms | 2 - 618 MB | NPU
68
- | Depth-Anything | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 34.715 ms | 5 - 659 MB | NPU
69
- | Depth-Anything | ONNX | float | Snapdragon® X2 Elite | 34.368 ms | 51 - 51 MB | NPU
70
- | Depth-Anything | ONNX | w8a16 | Snapdragon® X Elite | 54.946 ms | 29 - 29 MB | NPU
71
- | Depth-Anything | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 40.821 ms | 3 - 1054 MB | NPU
72
- | Depth-Anything | ONNX | w8a16 | Qualcomm® QCS6490 | 3258.093 ms | 160 - 164 MB | CPU
73
- | Depth-Anything | ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 55.751 ms | 0 - 38 MB | NPU
74
- | Depth-Anything | ONNX | w8a16 | Qualcomm® QCS9075 | 79.431 ms | 2 - 5 MB | NPU
75
- | Depth-Anything | ONNX | w8a16 | Qualcomm® QCM6690 | 1685.244 ms | 155 - 170 MB | CPU
76
- | Depth-Anything | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 32.257 ms | 3 - 838 MB | NPU
77
- | Depth-Anything | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 1636.432 ms | 157 - 172 MB | CPU
78
- | Depth-Anything | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 26.864 ms | 3 - 575 MB | NPU
79
- | Depth-Anything | ONNX | w8a16 | Snapdragon® X2 Elite | 28.445 ms | 30 - 30 MB | NPU
80
- | Depth-Anything | QNN_DLC | float | Snapdragon® X Elite | 76.386 ms | 3 - 3 MB | NPU
81
- | Depth-Anything | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 54.654 ms | 1 - 663 MB | NPU
82
- | Depth-Anything | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 176.285 ms | 2 - 493 MB | NPU
83
- | Depth-Anything | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 74.439 ms | 3 - 569 MB | NPU
84
- | Depth-Anything | QNN_DLC | float | Qualcomm® SA8775P | 80.749 ms | 1 - 496 MB | NPU
85
- | Depth-Anything | QNN_DLC | float | Qualcomm® QCS9075 | 97.156 ms | 3 - 9 MB | NPU
86
- | Depth-Anything | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 118.968 ms | 1 - 610 MB | NPU
87
- | Depth-Anything | QNN_DLC | float | Qualcomm® SA7255P | 176.285 ms | 2 - 493 MB | NPU
88
- | Depth-Anything | QNN_DLC | float | Qualcomm® SA8295P | 104.535 ms | 0 - 521 MB | NPU
89
- | Depth-Anything | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 40.88 ms | 2 - 493 MB | NPU
90
- | Depth-Anything | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 36.484 ms | 2 - 504 MB | NPU
91
- | Depth-Anything | QNN_DLC | float | Snapdragon® X2 Elite | 34.232 ms | 3 - 3 MB | NPU
92
- | Depth-Anything | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 50.38 ms | 1 - 591 MB | NPU
93
- | Depth-Anything | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 176.569 ms | 1 - 445 MB | NPU
94
- | Depth-Anything | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 72.423 ms | 1 - 12 MB | NPU
95
- | Depth-Anything | TFLITE | float | Qualcomm® SA8775P | 78.958 ms | 1 - 444 MB | NPU
96
- | Depth-Anything | TFLITE | float | Qualcomm® QCS9075 | 89.409 ms | 0 - 57 MB | NPU
97
- | Depth-Anything | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 105.705 ms | 1 - 580 MB | NPU
98
- | Depth-Anything | TFLITE | float | Qualcomm® SA7255P | 176.569 ms | 1 - 445 MB | NPU
99
- | Depth-Anything | TFLITE | float | Qualcomm® SA8295P | 101.125 ms | 1 - 435 MB | NPU
100
- | Depth-Anything | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 37.742 ms | 1 - 459 MB | NPU
101
- | Depth-Anything | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 31.437 ms | 1 - 466 MB | NPU
102
 
103
  ## License
104
  * The license for the original implementation of Depth-Anything can be found
 
14
  Depth Anything is designed for estimating depth at each point in an image.
15
 
16
  This is based on the implementation of Depth-Anything found [here](https://github.com/huggingface/transformers/tree/main/src/transformers/models/depth_anything).
17
+ 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/qai_hub_models/models/depth_anything) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
18
 
19
  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.
20
 
 
27
 
28
  | Runtime | Precision | Chipset | SDK Versions | Download |
29
  |---|---|---|---|---|
30
+ | ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.48.0/depth_anything-onnx-float.zip)
31
+ | ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.48.0/depth_anything-onnx-w8a16.zip)
32
+ | QNN_DLC | float | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.48.0/depth_anything-qnn_dlc-float.zip)
33
+ | QNN_DLC | w8a16 | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.48.0/depth_anything-qnn_dlc-w8a16.zip)
34
+ | TFLITE | float | Universal | QAIRT 2.43, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/depth_anything/releases/v0.48.0/depth_anything-tflite-float.zip)
35
 
36
  For more device-specific assets and performance metrics, visit **[Depth-Anything on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/depth_anything)**.
37
 
38
 
39
  ### Option 2: Export with Custom Configurations
40
 
41
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/depth_anything) Python library to compile and export the model with your own:
42
  - Custom weights (e.g., fine-tuned checkpoints)
43
  - Custom input shapes
44
  - Target device and runtime configurations
45
 
46
  This option is ideal if you need to customize the model beyond the default configuration provided here.
47
 
48
+ See our repository for [Depth-Anything on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/depth_anything) for usage instructions.
49
 
50
  ## Model Details
51
 
 
60
  ## Performance Summary
61
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
62
  |---|---|---|---|---|---|---
63
+ | Depth-Anything | ONNX | float | Snapdragon® X2 Elite | 34.426 ms | 51 - 51 MB | NPU
64
+ | Depth-Anything | ONNX | float | Snapdragon® X Elite | 79.451 ms | 50 - 50 MB | NPU
65
+ | Depth-Anything | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 56.069 ms | 0 - 864 MB | NPU
66
+ | Depth-Anything | ONNX | float | Qualcomm® QCS8550 (Proxy) | 76.065 ms | 0 - 74 MB | NPU
67
+ | Depth-Anything | ONNX | float | Qualcomm® QCS9075 | 98.771 ms | 3 - 9 MB | NPU
68
+ | Depth-Anything | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 41.536 ms | 2 - 657 MB | NPU
69
+ | Depth-Anything | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 34.71 ms | 5 - 663 MB | NPU
70
+ | Depth-Anything | ONNX | w8a16 | Snapdragon® X2 Elite | 26.345 ms | 30 - 30 MB | NPU
71
+ | Depth-Anything | ONNX | w8a16 | Snapdragon® X Elite | 54.994 ms | 29 - 29 MB | NPU
72
+ | Depth-Anything | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 40.976 ms | 3 - 1048 MB | NPU
73
+ | Depth-Anything | ONNX | w8a16 | Qualcomm® QCS6490 | 3284.497 ms | 160 - 165 MB | CPU
74
+ | Depth-Anything | ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 55.754 ms | 0 - 39 MB | NPU
75
+ | Depth-Anything | ONNX | w8a16 | Qualcomm® QCS9075 | 79.221 ms | 2 - 5 MB | NPU
76
+ | Depth-Anything | ONNX | w8a16 | Qualcomm® QCM6690 | 1689.741 ms | 159 - 175 MB | CPU
77
+ | Depth-Anything | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 32.169 ms | 3 - 835 MB | NPU
78
+ | Depth-Anything | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 1619.919 ms | 155 - 172 MB | CPU
79
+ | Depth-Anything | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 27.36 ms | 3 - 575 MB | NPU
80
+ | Depth-Anything | QNN_DLC | float | Snapdragon® X2 Elite | 34.452 ms | 3 - 3 MB | NPU
81
+ | Depth-Anything | QNN_DLC | float | Snapdragon® X Elite | 76.381 ms | 3 - 3 MB | NPU
82
+ | Depth-Anything | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 55.105 ms | 3 - 663 MB | NPU
83
+ | Depth-Anything | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 177.714 ms | 1 - 492 MB | NPU
84
+ | Depth-Anything | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 74.559 ms | 3 - 469 MB | NPU
85
+ | Depth-Anything | QNN_DLC | float | Qualcomm® SA8775P | 80.688 ms | 1 - 496 MB | NPU
86
+ | Depth-Anything | QNN_DLC | float | Qualcomm® QCS9075 | 96.743 ms | 5 - 11 MB | NPU
87
+ | Depth-Anything | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 118.74 ms | 1 - 609 MB | NPU
88
+ | Depth-Anything | QNN_DLC | float | Qualcomm® SA7255P | 177.714 ms | 1 - 492 MB | NPU
89
+ | Depth-Anything | QNN_DLC | float | Qualcomm® SA8295P | 104.569 ms | 0 - 521 MB | NPU
90
+ | Depth-Anything | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 40.816 ms | 3 - 495 MB | NPU
91
+ | Depth-Anything | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 36.061 ms | 3 - 505 MB | NPU
92
+ | Depth-Anything | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 50.593 ms | 0 - 600 MB | NPU
93
+ | Depth-Anything | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 176.552 ms | 1 - 445 MB | NPU
94
+ | Depth-Anything | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 72.366 ms | 1 - 12 MB | NPU
95
+ | Depth-Anything | TFLITE | float | Qualcomm® SA8775P | 377.253 ms | 1 - 445 MB | NPU
96
+ | Depth-Anything | TFLITE | float | Qualcomm® QCS9075 | 89.539 ms | 1 - 58 MB | NPU
97
+ | Depth-Anything | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 104.9 ms | 2 - 584 MB | NPU
98
+ | Depth-Anything | TFLITE | float | Qualcomm® SA7255P | 176.552 ms | 1 - 445 MB | NPU
99
+ | Depth-Anything | TFLITE | float | Qualcomm® SA8295P | 101.142 ms | 1 - 436 MB | NPU
100
+ | Depth-Anything | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 37.755 ms | 1 - 459 MB | NPU
101
+ | Depth-Anything | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 31.396 ms | 1 - 466 MB | NPU
102
 
103
  ## License
104
  * The license for the original implementation of Depth-Anything can be found