Hugging Face's logo

qualcomm
/
FFNet-54S-Quantized

Image Segmentation
PyTorch
TF Lite
ONNX
quantized
real_time
android
Model card Files Community
qaihm-bot commited on
Commit
b52195d
·
verified ·
1 Parent(s): 2194c35

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +23 -23
README.md CHANGED
@@ -37,36 +37,35 @@ More details on model performance across various devices, can be found
37
 
38
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
39
  |---|---|---|---|---|---|---|---|---|
40
- | FFNet-54S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 11.013 ms | 1 - 14 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
41
- | FFNet-54S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 14.203 ms | 6 - 49 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
42
- | FFNet-54S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 8.067 ms | 0 - 39 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
43
- | FFNet-54S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 9.961 ms | 6 - 55 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
44
- | FFNet-54S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 6.865 ms | 1 - 34 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
45
- | FFNet-54S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 9.524 ms | 6 - 51 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
46
- | FFNet-54S-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 44.715 ms | 1 - 41 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
47
- | FFNet-54S-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE | 320.173 ms | 1 - 3 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
48
- | FFNet-54S-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 10.981 ms | 1 - 17 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
49
- | FFNet-54S-Quantized | SA7255P ADP | SA7255P | TFLITE | 127.366 ms | 0 - 27 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
50
- | FFNet-54S-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 11.037 ms | 1 - 11 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
51
- | FFNet-54S-Quantized | SA8295P ADP | SA8295P | TFLITE | 16.583 ms | 1 - 35 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
52
- | FFNet-54S-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 11.063 ms | 1 - 14 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
53
- | FFNet-54S-Quantized | SA8775P ADP | SA8775P | TFLITE | 14.191 ms | 1 - 28 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
54
- | FFNet-54S-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 12.667 ms | 1 - 36 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
55
- | FFNet-54S-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 14.588 ms | 15 - 15 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
56
 
57
 
58
 
59
 
60
  ## Installation
61
 
62
- This model can be installed as a Python package via pip.
63
 
 
64
  ```bash
65
- pip install "qai-hub-models[ffnet_54s_quantized]"
66
  ```
67
 
68
 
69
-
70
  ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
71
 
72
  Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
@@ -117,8 +116,8 @@ Profiling Results
117
  FFNet-54S-Quantized
118
  Device : Samsung Galaxy S23 (13)
119
  Runtime : TFLITE
120
- Estimated inference time (ms) : 11.0
121
- Estimated peak memory usage (MB): [1, 14]
122
  Total # Ops : 117
123
  Compute Unit(s) : NPU (117 ops)
124
  ```
@@ -145,7 +144,7 @@ from qai_hub_models.models.ffnet_54s_quantized import Model
145
  torch_model = Model.from_pretrained()
146
 
147
  # Device
148
- device = hub.Device("Samsung Galaxy S23")
149
 
150
  # Trace model
151
  input_shape = torch_model.get_input_spec()
@@ -237,7 +236,8 @@ Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
237
 
238
 
239
  ## License
240
- * The license for the original implementation of FFNet-54S-Quantized can be found [here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE).
 
241
  * 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)
242
 
243
 
 
37
 
38
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
39
  |---|---|---|---|---|---|---|---|---|
40
+ | FFNet-54S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 11.134 ms | 1 - 12 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
41
+ | FFNet-54S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 14.134 ms | 6 - 33 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
42
+ | FFNet-54S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 8.062 ms | 1 - 35 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
43
+ | FFNet-54S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 10.329 ms | 6 - 56 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
44
+ | FFNet-54S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 8.24 ms | 0 - 35 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
45
+ | FFNet-54S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 9.412 ms | 8 - 53 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
46
+ | FFNet-54S-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 44.939 ms | 1 - 38 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
47
+ | FFNet-54S-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE | 321.494 ms | 1 - 4 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
48
+ | FFNet-54S-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 11.195 ms | 1 - 14 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
49
+ | FFNet-54S-Quantized | SA7255P ADP | SA7255P | TFLITE | 127.313 ms | 1 - 28 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
50
+ | FFNet-54S-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 11.113 ms | 1 - 19 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
51
+ | FFNet-54S-Quantized | SA8295P ADP | SA8295P | TFLITE | 16.588 ms | 1 - 35 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
52
+ | FFNet-54S-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 11.119 ms | 1 - 14 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
53
+ | FFNet-54S-Quantized | SA8775P ADP | SA8775P | TFLITE | 14.185 ms | 0 - 27 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
54
+ | FFNet-54S-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 12.748 ms | 1 - 36 MB | INT8 | NPU | [FFNet-54S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.tflite) |
55
+ | FFNet-54S-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 14.671 ms | 14 - 14 MB | INT8 | NPU | [FFNet-54S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-54S-Quantized/blob/main/FFNet-54S-Quantized.onnx) |
56
 
57
 
58
 
59
 
60
  ## Installation
61
 
 
62
 
63
+ Install the package via pip:
64
  ```bash
65
+ pip install "qai-hub-models[ffnet-54s-quantized]"
66
  ```
67
 
68
 
 
69
  ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
70
 
71
  Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
 
116
  FFNet-54S-Quantized
117
  Device : Samsung Galaxy S23 (13)
118
  Runtime : TFLITE
119
+ Estimated inference time (ms) : 11.1
120
+ Estimated peak memory usage (MB): [1, 12]
121
  Total # Ops : 117
122
  Compute Unit(s) : NPU (117 ops)
123
  ```
 
144
  torch_model = Model.from_pretrained()
145
 
146
  # Device
147
+ device = hub.Device("Samsung Galaxy S24")
148
 
149
  # Trace model
150
  input_shape = torch_model.get_input_spec()
 
236
 
237
 
238
  ## License
239
+ * The license for the original implementation of FFNet-54S-Quantized can be found
240
+ [here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE).
241
  * 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)
242
 
243