Hugging Face's logo

qualcomm
/
Depth-Anything-V2

Depth Estimation
PyTorch
TF Lite
ONNX
android
Model card Files Community
1
qaihm-bot commited on
Commit
9053913
·
verified ·
1 Parent(s): 152b254

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +34 -33
README.md CHANGED
@@ -34,44 +34,44 @@ More details on model performance across various devices, can be found
34
 
35
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
36
  |---|---|---|---|---|---|---|---|---|
37
- | Depth-Anything-V2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 320.265 ms | 2 - 52 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
38
- | Depth-Anything-V2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 378.409 ms | 3 - 83 MB | FP16 | NPU | [Depth-Anything-V2.so](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.so) |
39
- | Depth-Anything-V2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 238.814 ms | 2 - 182 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
40
- | Depth-Anything-V2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 250.977 ms | 0 - 244 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
41
- | Depth-Anything-V2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 287.329 ms | 76 - 329 MB | FP16 | NPU | [Depth-Anything-V2.so](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.so) |
42
- | Depth-Anything-V2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 187.475 ms | 5 - 279 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
43
- | Depth-Anything-V2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 209.94 ms | 0 - 270 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
44
- | Depth-Anything-V2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 274.456 ms | 0 - 280 MB | FP16 | NPU | Use Export Script |
45
- | Depth-Anything-V2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 152.637 ms | 5 - 296 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
46
- | Depth-Anything-V2 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 323.036 ms | 1 - 59 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
47
- | Depth-Anything-V2 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 227.663 ms | 3 - 6 MB | FP16 | NPU | Use Export Script |
48
- | Depth-Anything-V2 | SA7255P ADP | SA7255P | TFLITE | 1138.83 ms | 0 - 269 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
49
- | Depth-Anything-V2 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 329.941 ms | 1 - 51 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
50
- | Depth-Anything-V2 | SA8255 (Proxy) | SA8255P Proxy | QNN | 231.691 ms | 3 - 6 MB | FP16 | NPU | Use Export Script |
51
- | Depth-Anything-V2 | SA8295P ADP | SA8295P | TFLITE | 388.59 ms | 1 - 273 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
52
- | Depth-Anything-V2 | SA8295P ADP | SA8295P | QNN | 280.45 ms | 0 - 15 MB | FP16 | NPU | Use Export Script |
53
- | Depth-Anything-V2 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 326.538 ms | 0 - 79 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
54
- | Depth-Anything-V2 | SA8650 (Proxy) | SA8650P Proxy | QNN | 238.408 ms | 3 - 5 MB | FP16 | NPU | Use Export Script |
55
- | Depth-Anything-V2 | SA8775P ADP | SA8775P | TFLITE | 368.884 ms | 1 - 269 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
56
- | Depth-Anything-V2 | SA8775P ADP | SA8775P | QNN | 264.135 ms | 0 - 10 MB | FP16 | NPU | Use Export Script |
57
- | Depth-Anything-V2 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 369.58 ms | 1 - 260 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
58
- | Depth-Anything-V2 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 426.914 ms | 0 - 274 MB | FP16 | NPU | Use Export Script |
59
- | Depth-Anything-V2 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 228.152 ms | 3 - 3 MB | FP16 | NPU | Use Export Script |
60
- | Depth-Anything-V2 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 272.007 ms | 63 - 63 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
 
61
 
62
 
63
 
64
 
65
  ## Installation
66
 
67
- This model can be installed as a Python package via pip.
68
 
 
69
  ```bash
70
- pip install "qai-hub-models[depth_anything_v2]"
71
  ```
72
 
73
 
74
-
75
  ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
76
 
77
  Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
@@ -122,10 +122,10 @@ Profiling Results
122
  Depth-Anything-V2
123
  Device : Samsung Galaxy S23 (13)
124
  Runtime : TFLITE
125
- Estimated inference time (ms) : 320.3
126
- Estimated peak memory usage (MB): [2, 52]
127
- Total # Ops : 635
128
- Compute Unit(s) : NPU (635 ops)
129
  ```
130
 
131
 
@@ -150,7 +150,7 @@ from qai_hub_models.models.depth_anything_v2 import Model
150
  torch_model = Model.from_pretrained()
151
 
152
  # Device
153
- device = hub.Device("Samsung Galaxy S23")
154
 
155
  # Trace model
156
  input_shape = torch_model.get_input_spec()
@@ -242,7 +242,8 @@ Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
242
 
243
 
244
  ## License
245
- * The license for the original implementation of Depth-Anything-V2 can be found [here](https://github.com/huggingface/transformers/blob/main/LICENSE).
 
246
  * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
247
 
248
 
 
34
 
35
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
36
  |---|---|---|---|---|---|---|---|---|
37
+ | Depth-Anything-V2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 318.111 ms | 1 - 52 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
38
+ | Depth-Anything-V2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 377.493 ms | 0 - 59 MB | FP16 | NPU | [Depth-Anything-V2.so](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.so) |
39
+ | Depth-Anything-V2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 240.797 ms | 2 - 183 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
40
+ | Depth-Anything-V2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 254.117 ms | 0 - 247 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
41
+ | Depth-Anything-V2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 285.848 ms | 3 - 258 MB | FP16 | NPU | [Depth-Anything-V2.so](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.so) |
42
+ | Depth-Anything-V2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 186.226 ms | 3 - 276 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
43
+ | Depth-Anything-V2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 236.634 ms | 1 - 271 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
44
+ | Depth-Anything-V2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 243.26 ms | 3 - 284 MB | FP16 | NPU | Use Export Script |
45
+ | Depth-Anything-V2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 162.442 ms | 5 - 296 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
46
+ | Depth-Anything-V2 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 319.226 ms | 0 - 55 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
47
+ | Depth-Anything-V2 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 230.372 ms | 3 - 6 MB | FP16 | NPU | Use Export Script |
48
+ | Depth-Anything-V2 | SA7255P ADP | SA7255P | TFLITE | 1140.64 ms | 0 - 268 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
49
+ | Depth-Anything-V2 | SA7255P ADP | SA7255P | QNN | 1005.392 ms | 2 - 12 MB | FP16 | NPU | Use Export Script |
50
+ | Depth-Anything-V2 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 318.769 ms | 0 - 56 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
51
+ | Depth-Anything-V2 | SA8255 (Proxy) | SA8255P Proxy | QNN | 229.192 ms | 3 - 6 MB | FP16 | NPU | Use Export Script |
52
+ | Depth-Anything-V2 | SA8295P ADP | SA8295P | TFLITE | 388.234 ms | 1 - 273 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
53
+ | Depth-Anything-V2 | SA8295P ADP | SA8295P | QNN | 279.872 ms | 0 - 14 MB | FP16 | NPU | Use Export Script |
54
+ | Depth-Anything-V2 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 318.705 ms | 1 - 51 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
55
+ | Depth-Anything-V2 | SA8650 (Proxy) | SA8650P Proxy | QNN | 230.211 ms | 3 - 6 MB | FP16 | NPU | Use Export Script |
56
+ | Depth-Anything-V2 | SA8775P ADP | SA8775P | TFLITE | 368.849 ms | 1 - 269 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
57
+ | Depth-Anything-V2 | SA8775P ADP | SA8775P | QNN | 263.84 ms | 2 - 12 MB | FP16 | NPU | Use Export Script |
58
+ | Depth-Anything-V2 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 386.22 ms | 1 - 261 MB | FP16 | NPU | [Depth-Anything-V2.tflite](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.tflite) |
59
+ | Depth-Anything-V2 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 473.143 ms | 0 - 271 MB | FP16 | NPU | Use Export Script |
60
+ | Depth-Anything-V2 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 207.477 ms | 3 - 3 MB | FP16 | NPU | Use Export Script |
61
+ | Depth-Anything-V2 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 272.82 ms | 63 - 63 MB | FP16 | NPU | [Depth-Anything-V2.onnx](https://huggingface.co/qualcomm/Depth-Anything-V2/blob/main/Depth-Anything-V2.onnx) |
62
 
63
 
64
 
65
 
66
  ## Installation
67
 
 
68
 
69
+ Install the package via pip:
70
  ```bash
71
+ pip install "qai-hub-models[depth-anything-v2]"
72
  ```
73
 
74
 
 
75
  ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
76
 
77
  Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
 
122
  Depth-Anything-V2
123
  Device : Samsung Galaxy S23 (13)
124
  Runtime : TFLITE
125
+ Estimated inference time (ms) : 318.1
126
+ Estimated peak memory usage (MB): [1, 52]
127
+ Total # Ops : 634
128
+ Compute Unit(s) : NPU (634 ops)
129
  ```
130
 
131
 
 
150
  torch_model = Model.from_pretrained()
151
 
152
  # Device
153
+ device = hub.Device("Samsung Galaxy S24")
154
 
155
  # Trace model
156
  input_shape = torch_model.get_input_spec()
 
242
 
243
 
244
  ## License
245
+ * The license for the original implementation of Depth-Anything-V2 can be found
246
+ [here](https://github.com/huggingface/transformers/blob/main/LICENSE).
247
  * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
248
 
249