Hugging Face's logo

qualcomm
/
MediaPipe-Selfie-Segmentation

Image Segmentation
PyTorch
LiteRT
ONNX
android
Model card Files Community
qaihm-bot commited on
Commit
9bd82e6
·
verified ·
1 Parent(s): 36e5056

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +40 -19
README.md CHANGED
@@ -14,7 +14,7 @@ tags:
14
 
15
  Light-weight model that segments a person from the background in square or landscape selfie and video conference imagery.
16
 
17
- This model is an implementation of MediaPipe-Selfie-Segmentation found [here](https://github.com/google/mediapipe/tree/master/mediapipe/modules/selfie_segmentation).
18
  This repository provides scripts to run MediaPipe-Selfie-Segmentation on Qualcomm® devices.
19
  More details on model performance across various devices, can be found
20
  [here](https://aihub.qualcomm.com/models/mediapipe_selfie).
@@ -31,15 +31,32 @@ More details on model performance across various devices, can be found
31
  - Model size: 454 KB
32
  - Number of output classes: 6
33
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
34
 
35
 
36
 
37
- | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
38
- | ---|---|---|---|---|---|---|---|
39
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 0.699 ms | 0 - 2 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite)
40
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 0.775 ms | 1 - 4 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.so](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.so)
41
-
42
-
43
 
44
  ## Installation
45
 
@@ -95,16 +112,16 @@ device. This script does the following:
95
  ```bash
96
  python -m qai_hub_models.models.mediapipe_selfie.export
97
  ```
98
-
99
  ```
100
- Profile Job summary of MediaPipe-Selfie-Segmentation
101
- --------------------------------------------------
102
- Device: Snapdragon X Elite CRD (11)
103
- Estimated Inference Time: 0.92 ms
104
- Estimated Peak Memory Range: 0.75-0.75 MB
105
- Compute Units: NPU (138) | Total (138)
106
-
107
-
 
108
  ```
109
 
110
 
@@ -203,15 +220,19 @@ provides instructions on how to use the `.so` shared library in an Android appl
203
  Get more details on MediaPipe-Selfie-Segmentation's performance across various devices [here](https://aihub.qualcomm.com/models/mediapipe_selfie).
204
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
205
 
 
206
  ## License
207
- - The license for the original implementation of MediaPipe-Selfie-Segmentation can be found
208
- [here](https://github.com/google/mediapipe/blob/master/LICENSE).
209
- - 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)
 
210
 
211
  ## References
212
  * [Image segmentation guide](https://developers.google.com/mediapipe/solutions/vision/image_segmenter/)
213
  * [Source Model Implementation](https://github.com/google/mediapipe/tree/master/mediapipe/modules/selfie_segmentation)
214
 
 
 
215
  ## Community
216
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
217
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
 
14
 
15
  Light-weight model that segments a person from the background in square or landscape selfie and video conference imagery.
16
 
17
+ This model is an implementation of MediaPipe-Selfie-Segmentation found [here]({source_repo}).
18
  This repository provides scripts to run MediaPipe-Selfie-Segmentation on Qualcomm® devices.
19
  More details on model performance across various devices, can be found
20
  [here](https://aihub.qualcomm.com/models/mediapipe_selfie).
 
31
  - Model size: 454 KB
32
  - Number of output classes: 6
33
 
34
+ | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
35
+ |---|---|---|---|---|---|---|---|---|
36
+ | MediaPipe-Selfie-Segmentation | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 0.698 ms | 0 - 2 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
37
+ | MediaPipe-Selfie-Segmentation | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 0.774 ms | 1 - 25 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.so](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.so) |
38
+ | MediaPipe-Selfie-Segmentation | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 1.32 ms | 0 - 3 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.onnx](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.onnx) |
39
+ | MediaPipe-Selfie-Segmentation | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 0.472 ms | 0 - 29 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
40
+ | MediaPipe-Selfie-Segmentation | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 0.525 ms | 0 - 13 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.so](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.so) |
41
+ | MediaPipe-Selfie-Segmentation | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 0.899 ms | 0 - 32 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.onnx](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.onnx) |
42
+ | MediaPipe-Selfie-Segmentation | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 0.696 ms | 0 - 1 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
43
+ | MediaPipe-Selfie-Segmentation | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 0.756 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
44
+ | MediaPipe-Selfie-Segmentation | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 0.698 ms | 0 - 2 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
45
+ | MediaPipe-Selfie-Segmentation | SA8255 (Proxy) | SA8255P Proxy | QNN | 0.755 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
46
+ | MediaPipe-Selfie-Segmentation | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 0.697 ms | 0 - 68 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
47
+ | MediaPipe-Selfie-Segmentation | SA8775 (Proxy) | SA8775P Proxy | QNN | 0.76 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
48
+ | MediaPipe-Selfie-Segmentation | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 0.704 ms | 0 - 2 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
49
+ | MediaPipe-Selfie-Segmentation | SA8650 (Proxy) | SA8650P Proxy | QNN | 0.763 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
50
+ | MediaPipe-Selfie-Segmentation | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 0.934 ms | 0 - 28 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
51
+ | MediaPipe-Selfie-Segmentation | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 0.995 ms | 1 - 16 MB | FP16 | NPU | Use Export Script |
52
+ | MediaPipe-Selfie-Segmentation | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 0.367 ms | 0 - 18 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.tflite](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.tflite) |
53
+ | MediaPipe-Selfie-Segmentation | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 0.518 ms | 0 - 10 MB | FP16 | NPU | Use Export Script |
54
+ | MediaPipe-Selfie-Segmentation | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 0.871 ms | 0 - 23 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.onnx](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.onnx) |
55
+ | MediaPipe-Selfie-Segmentation | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 0.908 ms | 1 - 1 MB | FP16 | NPU | Use Export Script |
56
+ | MediaPipe-Selfie-Segmentation | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1.367 ms | 2 - 2 MB | FP16 | NPU | [MediaPipe-Selfie-Segmentation.onnx](https://huggingface.co/qualcomm/MediaPipe-Selfie-Segmentation/blob/main/MediaPipe-Selfie-Segmentation.onnx) |
57
 
58
 
59
 
 
 
 
 
 
 
60
 
61
  ## Installation
62
 
 
112
  ```bash
113
  python -m qai_hub_models.models.mediapipe_selfie.export
114
  ```
 
115
  ```
116
+ Profiling Results
117
+ ------------------------------------------------------------
118
+ MediaPipe-Selfie-Segmentation
119
+ Device : Samsung Galaxy S23 (13)
120
+ Runtime : TFLITE
121
+ Estimated inference time (ms) : 0.7
122
+ Estimated peak memory usage (MB): [0, 2]
123
+ Total # Ops : 118
124
+ Compute Unit(s) : NPU (118 ops)
125
  ```
126
 
127
 
 
220
  Get more details on MediaPipe-Selfie-Segmentation's performance across various devices [here](https://aihub.qualcomm.com/models/mediapipe_selfie).
221
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
222
 
223
+
224
  ## License
225
+ * The license for the original implementation of MediaPipe-Selfie-Segmentation can be found [here](https://github.com/google/mediapipe/blob/master/LICENSE).
226
+ * 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)
227
+
228
+
229
 
230
  ## References
231
  * [Image segmentation guide](https://developers.google.com/mediapipe/solutions/vision/image_segmenter/)
232
  * [Source Model Implementation](https://github.com/google/mediapipe/tree/master/mediapipe/modules/selfie_segmentation)
233
 
234
+
235
+
236
  ## Community
237
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
238
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).