Hugging Face's logo

qualcomm
/
FCN-ResNet50

Image Segmentation
PyTorch
LiteRT
android
Model card Files
xet
Community
bhushans commited on
Commit
4e17686
·
verified ·
1 Parent(s): 4c955fc

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +44 -26
README.md CHANGED
@@ -35,29 +35,31 @@ More details on model performance across various devices, can be found
35
 
36
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
- | FCN-ResNet50 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 62.203 ms | 0 - 2 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
39
- | FCN-ResNet50 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 62.537 ms | 3 - 27 MB | FP16 | NPU | [FCN-ResNet50.so](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.so) |
40
- | FCN-ResNet50 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 58.725 ms | 0 - 79 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
41
- | FCN-ResNet50 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 47.988 ms | 0 - 288 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
42
- | FCN-ResNet50 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 47.738 ms | 0 - 147 MB | FP16 | NPU | [FCN-ResNet50.so](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.so) |
43
- | FCN-ResNet50 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 46.355 ms | 0 - 304 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
44
- | FCN-ResNet50 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 47.855 ms | 0 - 204 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
45
- | FCN-ResNet50 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 39.07 ms | 3 - 147 MB | FP16 | NPU | Use Export Script |
46
- | FCN-ResNet50 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 45.793 ms | 2 - 222 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
47
- | FCN-ResNet50 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 61.841 ms | 0 - 2 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
48
- | FCN-ResNet50 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 58.931 ms | 3 - 5 MB | FP16 | NPU | Use Export Script |
49
- | FCN-ResNet50 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 62.177 ms | 0 - 3 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
50
- | FCN-ResNet50 | SA8255 (Proxy) | SA8255P Proxy | QNN | 59.156 ms | 3 - 5 MB | FP16 | NPU | Use Export Script |
51
- | FCN-ResNet50 | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 61.826 ms | 0 - 2 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
52
- | FCN-ResNet50 | SA8775 (Proxy) | SA8775P Proxy | QNN | 59.171 ms | 3 - 5 MB | FP16 | NPU | Use Export Script |
53
- | FCN-ResNet50 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 63.41 ms | 0 - 2 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
54
- | FCN-ResNet50 | SA8650 (Proxy) | SA8650P Proxy | QNN | 59.302 ms | 3 - 5 MB | FP16 | NPU | Use Export Script |
55
- | FCN-ResNet50 | SA8295P ADP | SA8295P | TFLITE | 101.22 ms | 0 - 66 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
56
- | FCN-ResNet50 | SA8295P ADP | SA8295P | QNN | 96.956 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
57
- | FCN-ResNet50 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 91.681 ms | 0 - 127 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
58
- | FCN-ResNet50 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 91.903 ms | 3 - 56 MB | FP16 | NPU | Use Export Script |
59
- | FCN-ResNet50 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 57.375 ms | 3 - 3 MB | FP16 | NPU | Use Export Script |
60
- | FCN-ResNet50 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 57.745 ms | 68 - 68 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
 
 
61
 
62
 
63
 
@@ -121,8 +123,8 @@ Profiling Results
121
  FCN-ResNet50
122
  Device : Samsung Galaxy S23 (13)
123
  Runtime : TFLITE
124
- Estimated inference time (ms) : 62.2
125
- Estimated peak memory usage (MB): [0, 2]
126
  Total # Ops : 88
127
  Compute Unit(s) : NPU (88 ops)
128
  ```
@@ -143,13 +145,29 @@ in memory using the `jit.trace` and then call the `submit_compile_job` API.
143
  import torch
144
 
145
  import qai_hub as hub
146
- from qai_hub_models.models.fcn_resnet50 import
147
 
148
  # Load the model
 
149
 
150
  # Device
151
  device = hub.Device("Samsung Galaxy S23")
152
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
153
 
154
  ```
155
 
 
35
 
36
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
+ | FCN-ResNet50 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 48.668 ms | 0 - 17 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
39
+ | FCN-ResNet50 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 48.277 ms | 3 - 21 MB | FP16 | NPU | [FCN-ResNet50.so](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.so) |
40
+ | FCN-ResNet50 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 43.588 ms | 0 - 79 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
41
+ | FCN-ResNet50 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 35.692 ms | 0 - 93 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
42
+ | FCN-ResNet50 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 35.984 ms | 3 - 95 MB | FP16 | NPU | [FCN-ResNet50.so](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.so) |
43
+ | FCN-ResNet50 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 33.635 ms | 0 - 217 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
44
+ | FCN-ResNet50 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 34.871 ms | 0 - 87 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
45
+ | FCN-ResNet50 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 29.555 ms | 3 - 91 MB | FP16 | NPU | Use Export Script |
46
+ | FCN-ResNet50 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 32.448 ms | 4 - 147 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
47
+ | FCN-ResNet50 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 48.109 ms | 0 - 18 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
48
+ | FCN-ResNet50 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 43.504 ms | 3 - 4 MB | FP16 | NPU | Use Export Script |
49
+ | FCN-ResNet50 | SA7255P ADP | SA7255P | TFLITE | 2068.721 ms | 0 - 86 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
50
+ | FCN-ResNet50 | SA7255P ADP | SA7255P | QNN | 2060.867 ms | 0 - 7 MB | FP16 | NPU | Use Export Script |
51
+ | FCN-ResNet50 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 48.503 ms | 0 - 17 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
52
+ | FCN-ResNet50 | SA8255 (Proxy) | SA8255P Proxy | QNN | 44.059 ms | 4 - 5 MB | FP16 | NPU | Use Export Script |
53
+ | FCN-ResNet50 | SA8295P ADP | SA8295P | TFLITE | 84.801 ms | 0 - 39 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
54
+ | FCN-ResNet50 | SA8295P ADP | SA8295P | QNN | 81.195 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
55
+ | FCN-ResNet50 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 48.852 ms | 0 - 19 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
56
+ | FCN-ResNet50 | SA8650 (Proxy) | SA8650P Proxy | QNN | 43.64 ms | 4 - 5 MB | FP16 | NPU | Use Export Script |
57
+ | FCN-ResNet50 | SA8775P ADP | SA8775P | TFLITE | 95.278 ms | 0 - 86 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
58
+ | FCN-ResNet50 | SA8775P ADP | SA8775P | QNN | 89.441 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
59
+ | FCN-ResNet50 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 78.386 ms | 0 - 46 MB | FP16 | NPU | [FCN-ResNet50.tflite](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.tflite) |
60
+ | FCN-ResNet50 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 75.75 ms | 3 - 48 MB | FP16 | NPU | Use Export Script |
61
+ | FCN-ResNet50 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 44.221 ms | 3 - 3 MB | FP16 | NPU | Use Export Script |
62
+ | FCN-ResNet50 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 45.093 ms | 63 - 63 MB | FP16 | NPU | [FCN-ResNet50.onnx](https://huggingface.co/qualcomm/FCN-ResNet50/blob/main/FCN-ResNet50.onnx) |
63
 
64
 
65
 
 
123
  FCN-ResNet50
124
  Device : Samsung Galaxy S23 (13)
125
  Runtime : TFLITE
126
+ Estimated inference time (ms) : 48.7
127
+ Estimated peak memory usage (MB): [0, 17]
128
  Total # Ops : 88
129
  Compute Unit(s) : NPU (88 ops)
130
  ```
 
145
  import torch
146
 
147
  import qai_hub as hub
148
+ from qai_hub_models.models.fcn_resnet50 import Model
149
 
150
  # Load the model
151
+ torch_model = Model.from_pretrained()
152
 
153
  # Device
154
  device = hub.Device("Samsung Galaxy S23")
155
 
156
+ # Trace model
157
+ input_shape = torch_model.get_input_spec()
158
+ sample_inputs = torch_model.sample_inputs()
159
+
160
+ pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
161
+
162
+ # Compile model on a specific device
163
+ compile_job = hub.submit_compile_job(
164
+ model=pt_model,
165
+ device=device,
166
+ input_specs=torch_model.get_input_spec(),
167
+ )
168
+
169
+ # Get target model to run on-device
170
+ target_model = compile_job.get_target_model()
171
 
172
  ```
173