Hugging Face's logo

qualcomm
/
MobileNet-v2

Image Classification
PyTorch
LiteRT
ONNX
backbone
real_time
android
Model card Files Community
bhushans commited on
Commit
4de1bc5
·
verified ·
1 Parent(s): e404074

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +46 -28
README.md CHANGED
@@ -1,6 +1,6 @@
1
  ---
2
  library_name: pytorch
3
- license: bsd-3-clause
4
  pipeline_tag: image-classification
5
  tags:
6
  - backbone
@@ -36,29 +36,31 @@ More details on model performance across various devices, can be found
36
 
37
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
38
  |---|---|---|---|---|---|---|---|---|
39
- | MobileNet-v2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 0.902 ms | 0 - 3 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
40
- | MobileNet-v2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 1.254 ms | 1 - 48 MB | FP16 | NPU | [MobileNet-v2.so](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.so) |
41
- | MobileNet-v2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 0.905 ms | 1 - 3 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
42
- | MobileNet-v2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 0.612 ms | 0 - 61 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
43
- | MobileNet-v2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 0.842 ms | 0 - 17 MB | FP16 | NPU | [MobileNet-v2.so](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.so) |
44
- | MobileNet-v2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 0.65 ms | 0 - 65 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
45
- | MobileNet-v2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 0.613 ms | 0 - 24 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
46
- | MobileNet-v2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 0.849 ms | 0 - 14 MB | FP16 | NPU | Use Export Script |
47
- | MobileNet-v2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 0.687 ms | 0 - 24 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
48
- | MobileNet-v2 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 0.901 ms | 0 - 2 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
49
- | MobileNet-v2 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 1.184 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
50
- | MobileNet-v2 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 0.904 ms | 0 - 177 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
51
- | MobileNet-v2 | SA8255 (Proxy) | SA8255P Proxy | QNN | 1.186 ms | 2 - 4 MB | FP16 | NPU | Use Export Script |
52
- | MobileNet-v2 | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 0.898 ms | 0 - 2 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
53
- | MobileNet-v2 | SA8775 (Proxy) | SA8775P Proxy | QNN | 1.192 ms | 3 - 4 MB | FP16 | NPU | Use Export Script |
54
- | MobileNet-v2 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 0.909 ms | 0 - 2 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
55
- | MobileNet-v2 | SA8650 (Proxy) | SA8650P Proxy | QNN | 1.194 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
56
- | MobileNet-v2 | SA8295P ADP | SA8295P | TFLITE | 1.513 ms | 0 - 23 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
57
- | MobileNet-v2 | SA8295P ADP | SA8295P | QNN | 2.093 ms | 0 - 5 MB | FP16 | NPU | Use Export Script |
58
- | MobileNet-v2 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 1.081 ms | 0 - 62 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
59
- | MobileNet-v2 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 1.442 ms | 1 - 19 MB | FP16 | NPU | Use Export Script |
60
- | MobileNet-v2 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 1.372 ms | 1 - 1 MB | FP16 | NPU | Use Export Script |
61
- | MobileNet-v2 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 0.977 ms | 9 - 9 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
 
 
62
 
63
 
64
 
@@ -123,9 +125,9 @@ MobileNet-v2
123
  Device : Samsung Galaxy S23 (13)
124
  Runtime : TFLITE
125
  Estimated inference time (ms) : 0.9
126
- Estimated peak memory usage (MB): [0, 3]
127
- Total # Ops : 72
128
- Compute Unit(s) : NPU (72 ops)
129
  ```
130
 
131
 
@@ -144,13 +146,29 @@ in memory using the `jit.trace` and then call the `submit_compile_job` API.
144
  import torch
145
 
146
  import qai_hub as hub
147
- from qai_hub_models.models.mobilenet_v2 import
148
 
149
  # Load the model
 
150
 
151
  # Device
152
  device = hub.Device("Samsung Galaxy S23")
153
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
154
 
155
  ```
156
 
 
1
  ---
2
  library_name: pytorch
3
+ license: apache-2.0
4
  pipeline_tag: image-classification
5
  tags:
6
  - backbone
 
36
 
37
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
38
  |---|---|---|---|---|---|---|---|---|
39
+ | MobileNet-v2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 0.88 ms | 0 - 29 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
40
+ | MobileNet-v2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 1.089 ms | 0 - 199 MB | FP16 | NPU | [MobileNet-v2.so](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.so) |
41
+ | MobileNet-v2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 0.905 ms | 0 - 2 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
42
+ | MobileNet-v2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 0.59 ms | 0 - 18 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
43
+ | MobileNet-v2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 0.725 ms | 0 - 16 MB | FP16 | NPU | [MobileNet-v2.so](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.so) |
44
+ | MobileNet-v2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 0.629 ms | 0 - 65 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
45
+ | MobileNet-v2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 0.59 ms | 0 - 12 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
46
+ | MobileNet-v2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 0.732 ms | 0 - 14 MB | FP16 | NPU | Use Export Script |
47
+ | MobileNet-v2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 0.658 ms | 0 - 24 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
48
+ | MobileNet-v2 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 0.872 ms | 0 - 153 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
49
+ | MobileNet-v2 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 1.035 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
50
+ | MobileNet-v2 | SA7255P ADP | SA7255P | TFLITE | 12.915 ms | 0 - 13 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
51
+ | MobileNet-v2 | SA7255P ADP | SA7255P | QNN | 13.295 ms | 1 - 6 MB | FP16 | NPU | Use Export Script |
52
+ | MobileNet-v2 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 0.883 ms | 0 - 71 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
53
+ | MobileNet-v2 | SA8255 (Proxy) | SA8255P Proxy | QNN | 1.038 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
54
+ | MobileNet-v2 | SA8295P ADP | SA8295P | TFLITE | 1.455 ms | 0 - 10 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
55
+ | MobileNet-v2 | SA8295P ADP | SA8295P | QNN | 1.73 ms | 1 - 6 MB | FP16 | NPU | Use Export Script |
56
+ | MobileNet-v2 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 0.879 ms | 0 - 203 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
57
+ | MobileNet-v2 | SA8650 (Proxy) | SA8650P Proxy | QNN | 1.036 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
58
+ | MobileNet-v2 | SA8775P ADP | SA8775P | TFLITE | 1.47 ms | 0 - 13 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
59
+ | MobileNet-v2 | SA8775P ADP | SA8775P | QNN | 1.872 ms | 1 - 6 MB | FP16 | NPU | Use Export Script |
60
+ | MobileNet-v2 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 1.053 ms | 0 - 16 MB | FP16 | NPU | [MobileNet-v2.tflite](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.tflite) |
61
+ | MobileNet-v2 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 1.276 ms | 1 - 18 MB | FP16 | NPU | Use Export Script |
62
+ | MobileNet-v2 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 1.21 ms | 1 - 1 MB | FP16 | NPU | Use Export Script |
63
+ | MobileNet-v2 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 0.939 ms | 8 - 8 MB | FP16 | NPU | [MobileNet-v2.onnx](https://huggingface.co/qualcomm/MobileNet-v2/blob/main/MobileNet-v2.onnx) |
64
 
65
 
66
 
 
125
  Device : Samsung Galaxy S23 (13)
126
  Runtime : TFLITE
127
  Estimated inference time (ms) : 0.9
128
+ Estimated peak memory usage (MB): [0, 29]
129
+ Total # Ops : 71
130
+ Compute Unit(s) : NPU (71 ops)
131
  ```
132
 
133
 
 
146
  import torch
147
 
148
  import qai_hub as hub
149
+ from qai_hub_models.models.mobilenet_v2 import Model
150
 
151
  # Load the model
152
+ torch_model = Model.from_pretrained()
153
 
154
  # Device
155
  device = hub.Device("Samsung Galaxy S23")
156
 
157
+ # Trace model
158
+ input_shape = torch_model.get_input_spec()
159
+ sample_inputs = torch_model.sample_inputs()
160
+
161
+ pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
162
+
163
+ # Compile model on a specific device
164
+ compile_job = hub.submit_compile_job(
165
+ model=pt_model,
166
+ device=device,
167
+ input_specs=torch_model.get_input_spec(),
168
+ )
169
+
170
+ # Get target model to run on-device
171
+ target_model = compile_job.get_target_model()
172
 
173
  ```
174