Hugging Face's logo

qualcomm
/
FFNet-54S

Image Segmentation
PyTorch
LiteRT
android
Model card Files Community
bhushans commited on
Commit
d67f551
·
verified ·
1 Parent(s): d6d7706

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +43 -26
README.md CHANGED
@@ -35,29 +35,30 @@ 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
- | FFNet-54S | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 19.705 ms | 2 - 4 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
39
- | FFNet-54S | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 20.103 ms | 24 - 48 MB | FP16 | NPU | [FFNet-54S.so](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.so) |
40
- | FFNet-54S | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 27.922 ms | 24 - 38 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
41
- | FFNet-54S | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 14.531 ms | 0 - 112 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
42
- | FFNet-54S | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 14.486 ms | 24 - 60 MB | FP16 | NPU | [FFNet-54S.so](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.so) |
43
- | FFNet-54S | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 21.1 ms | 29 - 167 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
44
- | FFNet-54S | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 14.005 ms | 1 - 48 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
45
- | FFNet-54S | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 14.044 ms | 24 - 59 MB | FP16 | NPU | Use Export Script |
46
- | FFNet-54S | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 22.09 ms | 28 - 78 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
47
- | FFNet-54S | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 19.757 ms | 2 - 5 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
48
- | FFNet-54S | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 19.382 ms | 24 - 25 MB | FP16 | NPU | Use Export Script |
49
- | FFNet-54S | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 19.681 ms | 2 - 4 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
50
- | FFNet-54S | SA8255 (Proxy) | SA8255P Proxy | QNN | 19.33 ms | 26 - 27 MB | FP16 | NPU | Use Export Script |
51
- | FFNet-54S | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 19.99 ms | 2 - 4 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
52
- | FFNet-54S | SA8775 (Proxy) | SA8775P Proxy | QNN | 19.296 ms | 26 - 27 MB | FP16 | NPU | Use Export Script |
53
- | FFNet-54S | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 19.839 ms | 2 - 4 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
54
- | FFNet-54S | SA8650 (Proxy) | SA8650P Proxy | QNN | 19.248 ms | 25 - 26 MB | FP16 | NPU | Use Export Script |
55
- | FFNet-54S | SA8295P ADP | SA8295P | TFLITE | 36.554 ms | 2 - 41 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
56
- | FFNet-54S | SA8295P ADP | SA8295P | QNN | 35.465 ms | 24 - 30 MB | FP16 | NPU | Use Export Script |
57
- | FFNet-54S | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 32.381 ms | 2 - 99 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
58
- | FFNet-54S | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 32.941 ms | 24 - 56 MB | FP16 | NPU | Use Export Script |
59
- | FFNet-54S | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 19.314 ms | 24 - 24 MB | FP16 | NPU | Use Export Script |
60
- | FFNet-54S | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 32.701 ms | 24 - 24 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
 
61
 
62
 
63
 
@@ -122,8 +123,8 @@ Profiling Results
122
  FFNet-54S
123
  Device : Samsung Galaxy S23 (13)
124
  Runtime : TFLITE
125
- Estimated inference time (ms) : 19.7
126
- Estimated peak memory usage (MB): [2, 4]
127
  Total # Ops : 113
128
  Compute Unit(s) : NPU (113 ops)
129
  ```
@@ -144,13 +145,29 @@ 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.ffnet_54s import
148
 
149
  # Load the model
 
150
 
151
  # Device
152
  device = hub.Device("Samsung Galaxy S23")
153
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
154
 
155
  ```
156
 
 
35
 
36
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
+ | FFNet-54S | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 19.845 ms | 2 - 28 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
39
+ | FFNet-54S | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 20.072 ms | 24 - 47 MB | FP16 | NPU | [FFNet-54S.so](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.so) |
40
+ | FFNet-54S | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 33.996 ms | 24 - 38 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
41
+ | FFNet-54S | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 14.403 ms | 2 - 40 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
42
+ | FFNet-54S | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 14.536 ms | 20 - 55 MB | FP16 | NPU | [FFNet-54S.so](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.so) |
43
+ | FFNet-54S | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 21.845 ms | 1 - 131 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
44
+ | FFNet-54S | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 14.04 ms | 1 - 37 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
45
+ | FFNet-54S | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 14.258 ms | 24 - 59 MB | FP16 | NPU | Use Export Script |
46
+ | FFNet-54S | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 22.078 ms | 29 - 82 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
47
+ | FFNet-54S | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 20.007 ms | 2 - 27 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
48
+ | FFNet-54S | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 19.007 ms | 24 - 25 MB | FP16 | NPU | Use Export Script |
49
+ | FFNet-54S | SA7255P ADP | SA7255P | TFLITE | 920.83 ms | 0 - 35 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
50
+ | FFNet-54S | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 19.775 ms | 2 - 28 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
51
+ | FFNet-54S | SA8255 (Proxy) | SA8255P Proxy | QNN | 19.115 ms | 25 - 27 MB | FP16 | NPU | Use Export Script |
52
+ | FFNet-54S | SA8295P ADP | SA8295P | TFLITE | 36.548 ms | 2 - 31 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
53
+ | FFNet-54S | SA8295P ADP | SA8295P | QNN | 35.494 ms | 24 - 30 MB | FP16 | NPU | Use Export Script |
54
+ | FFNet-54S | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 20.094 ms | 2 - 30 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
55
+ | FFNet-54S | SA8650 (Proxy) | SA8650P Proxy | QNN | 19.38 ms | 24 - 25 MB | FP16 | NPU | Use Export Script |
56
+ | FFNet-54S | SA8775P ADP | SA8775P | TFLITE | 41.163 ms | 1 - 35 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
57
+ | FFNet-54S | SA8775P ADP | SA8775P | QNN | 40.517 ms | 24 - 30 MB | FP16 | NPU | Use Export Script |
58
+ | FFNet-54S | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 32.828 ms | 2 - 38 MB | FP16 | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
59
+ | FFNet-54S | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 32.701 ms | 24 - 56 MB | FP16 | NPU | Use Export Script |
60
+ | FFNet-54S | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 19.504 ms | 24 - 24 MB | FP16 | NPU | Use Export Script |
61
+ | FFNet-54S | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 33.04 ms | 24 - 24 MB | FP16 | NPU | [FFNet-54S.onnx](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx) |
62
 
63
 
64
 
 
123
  FFNet-54S
124
  Device : Samsung Galaxy S23 (13)
125
  Runtime : TFLITE
126
+ Estimated inference time (ms) : 19.8
127
+ Estimated peak memory usage (MB): [2, 28]
128
  Total # Ops : 113
129
  Compute Unit(s) : NPU (113 ops)
130
  ```
 
145
  import torch
146
 
147
  import qai_hub as hub
148
+ from qai_hub_models.models.ffnet_54s 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