Hugging Face's logo

qualcomm
/
FFNet-78S-LowRes

Image Segmentation
PyTorch
real_time
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
49fee9a
·
verified ·
1 Parent(s): b7bc515

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +40 -19
README.md CHANGED
@@ -17,7 +17,7 @@ tags:
17
 
18
  FFNet-78S-LowRes is a "fuss-free network" that segments street scene images with per-pixel classes like road, sidewalk, and pedestrian. Trained on the Cityscapes dataset.
19
 
20
- This model is an implementation of FFNet-78S-LowRes found [here](https://github.com/Qualcomm-AI-research/FFNet).
21
  This repository provides scripts to run FFNet-78S-LowRes on Qualcomm® devices.
22
  More details on model performance across various devices, can be found
23
  [here](https://aihub.qualcomm.com/models/ffnet_78s_lowres).
@@ -33,15 +33,32 @@ More details on model performance across various devices, can be found
33
  - Model size: 102 MB
34
  - Number of output classes: 19
35
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36
 
37
 
38
 
39
- | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
40
- | ---|---|---|---|---|---|---|---|
41
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 8.311 ms | 1 - 3 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite)
42
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 8.366 ms | 6 - 34 MB | FP16 | NPU | [FFNet-78S-LowRes.so](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.so)
43
-
44
-
45
 
46
  ## Installation
47
 
@@ -97,16 +114,16 @@ device. This script does the following:
97
  ```bash
98
  python -m qai_hub_models.models.ffnet_78s_lowres.export
99
  ```
100
-
101
  ```
102
- Profile Job summary of FFNet-78S-LowRes
103
- --------------------------------------------------
104
- Device: Snapdragon X Elite CRD (11)
105
- Estimated Inference Time: 8.19 ms
106
- Estimated Peak Memory Range: 6.01-6.01 MB
107
- Compute Units: NPU (236) | Total (236)
108
-
109
-
 
110
  ```
111
 
112
 
@@ -205,15 +222,19 @@ provides instructions on how to use the `.so` shared library in an Android appl
205
  Get more details on FFNet-78S-LowRes's performance across various devices [here](https://aihub.qualcomm.com/models/ffnet_78s_lowres).
206
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
207
 
 
208
  ## License
209
- - The license for the original implementation of FFNet-78S-LowRes can be found
210
- [here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE).
211
- - 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)
 
212
 
213
  ## References
214
  * [Simple and Efficient Architectures for Semantic Segmentation](https://arxiv.org/abs/2206.08236)
215
  * [Source Model Implementation](https://github.com/Qualcomm-AI-research/FFNet)
216
 
 
 
217
  ## Community
218
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
219
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
 
17
 
18
  FFNet-78S-LowRes is a "fuss-free network" that segments street scene images with per-pixel classes like road, sidewalk, and pedestrian. Trained on the Cityscapes dataset.
19
 
20
+ This model is an implementation of FFNet-78S-LowRes found [here]({source_repo}).
21
  This repository provides scripts to run FFNet-78S-LowRes on Qualcomm® devices.
22
  More details on model performance across various devices, can be found
23
  [here](https://aihub.qualcomm.com/models/ffnet_78s_lowres).
 
33
  - Model size: 102 MB
34
  - Number of output classes: 19
35
 
36
+ | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
+ |---|---|---|---|---|---|---|---|---|
38
+ | FFNet-78S-LowRes | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 8.33 ms | 1 - 3 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
39
+ | FFNet-78S-LowRes | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 8.398 ms | 6 - 28 MB | FP16 | NPU | [FFNet-78S-LowRes.so](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.so) |
40
+ | FFNet-78S-LowRes | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 8.025 ms | 6 - 9 MB | FP16 | NPU | [FFNet-78S-LowRes.onnx](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx) |
41
+ | FFNet-78S-LowRes | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 6.595 ms | 1 - 61 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
42
+ | FFNet-78S-LowRes | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 7.132 ms | 6 - 32 MB | FP16 | NPU | [FFNet-78S-LowRes.so](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.so) |
43
+ | FFNet-78S-LowRes | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 6.923 ms | 2 - 82 MB | FP16 | NPU | [FFNet-78S-LowRes.onnx](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx) |
44
+ | FFNet-78S-LowRes | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 8.303 ms | 0 - 20 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
45
+ | FFNet-78S-LowRes | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 7.635 ms | 6 - 7 MB | FP16 | NPU | Use Export Script |
46
+ | FFNet-78S-LowRes | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 8.344 ms | 0 - 2 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
47
+ | FFNet-78S-LowRes | SA8255 (Proxy) | SA8255P Proxy | QNN | 7.638 ms | 6 - 8 MB | FP16 | NPU | Use Export Script |
48
+ | FFNet-78S-LowRes | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 8.181 ms | 0 - 4 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
49
+ | FFNet-78S-LowRes | SA8775 (Proxy) | SA8775P Proxy | QNN | 7.735 ms | 6 - 7 MB | FP16 | NPU | Use Export Script |
50
+ | FFNet-78S-LowRes | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 8.18 ms | 0 - 9 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
51
+ | FFNet-78S-LowRes | SA8650 (Proxy) | SA8650P Proxy | QNN | 7.747 ms | 6 - 7 MB | FP16 | NPU | Use Export Script |
52
+ | FFNet-78S-LowRes | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 11.977 ms | 1 - 54 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
53
+ | FFNet-78S-LowRes | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 12.509 ms | 6 - 25 MB | FP16 | NPU | Use Export Script |
54
+ | FFNet-78S-LowRes | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 5.656 ms | 0 - 29 MB | FP16 | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
55
+ | FFNet-78S-LowRes | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 5.912 ms | 6 - 26 MB | FP16 | NPU | Use Export Script |
56
+ | FFNet-78S-LowRes | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 4.997 ms | 7 - 50 MB | FP16 | NPU | [FFNet-78S-LowRes.onnx](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx) |
57
+ | FFNet-78S-LowRes | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 8.198 ms | 6 - 6 MB | FP16 | NPU | Use Export Script |
58
+ | FFNet-78S-LowRes | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 8.793 ms | 48 - 48 MB | FP16 | NPU | [FFNet-78S-LowRes.onnx](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx) |
59
 
60
 
61
 
 
 
 
 
 
 
62
 
63
  ## Installation
64
 
 
114
  ```bash
115
  python -m qai_hub_models.models.ffnet_78s_lowres.export
116
  ```
 
117
  ```
118
+ Profiling Results
119
+ ------------------------------------------------------------
120
+ FFNet-78S-LowRes
121
+ Device : Samsung Galaxy S23 (13)
122
+ Runtime : TFLITE
123
+ Estimated inference time (ms) : 8.3
124
+ Estimated peak memory usage (MB): [1, 3]
125
+ Total # Ops : 149
126
+ Compute Unit(s) : NPU (149 ops)
127
  ```
128
 
129
 
 
222
  Get more details on FFNet-78S-LowRes's performance across various devices [here](https://aihub.qualcomm.com/models/ffnet_78s_lowres).
223
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
224
 
225
+
226
  ## License
227
+ * The license for the original implementation of FFNet-78S-LowRes can be found [here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE).
228
+ * 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)
229
+
230
+
231
 
232
  ## References
233
  * [Simple and Efficient Architectures for Semantic Segmentation](https://arxiv.org/abs/2206.08236)
234
  * [Source Model Implementation](https://github.com/Qualcomm-AI-research/FFNet)
235
 
236
+
237
+
238
  ## Community
239
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
240
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).