Hugging Face's logo

qualcomm
/
Posenet-Mobilenet-Quantized

Keypoint Detection
PyTorch
LiteRT
quantized
android
Model card Files Community
qaihm-bot commited on
Commit
389d87b
·
verified ·
1 Parent(s): 4a8553e

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +39 -19
README.md CHANGED
@@ -17,7 +17,7 @@ tags:
17
 
18
  Posenet performs pose estimation on human images.
19
 
20
- This model is an implementation of Posenet-Mobilenet-Quantized found [here](https://github.com/rwightman/posenet-pytorch).
21
  This repository provides scripts to run Posenet-Mobilenet-Quantized on Qualcomm® devices.
22
  More details on model performance across various devices, can be found
23
  [here](https://aihub.qualcomm.com/models/posenet_mobilenet_quantized).
@@ -32,15 +32,31 @@ More details on model performance across various devices, can be found
32
  - Number of parameters: 3.31M
33
  - Model size: 3.47 MB
34
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
35
 
36
 
37
 
38
- | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
39
- | ---|---|---|---|---|---|---|---|
40
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 0.56 ms | 0 - 62 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite)
41
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 0.644 ms | 0 - 12 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.so](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.so)
42
-
43
-
44
 
45
  ## Installation
46
 
@@ -95,16 +111,16 @@ device. This script does the following:
95
  ```bash
96
  python -m qai_hub_models.models.posenet_mobilenet_quantized.export
97
  ```
98
-
99
  ```
100
- Profile Job summary of Posenet-Mobilenet-Quantized
101
- --------------------------------------------------
102
- Device: Snapdragon X Elite CRD (11)
103
- Estimated Inference Time: 0.67 ms
104
- Estimated Peak Memory Range: 0.38-0.38 MB
105
- Compute Units: NPU (42) | Total (42)
106
-
107
-
 
108
  ```
109
 
110
 
@@ -143,15 +159,19 @@ provides instructions on how to use the `.so` shared library in an Android appl
143
  Get more details on Posenet-Mobilenet-Quantized's performance across various devices [here](https://aihub.qualcomm.com/models/posenet_mobilenet_quantized).
144
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
145
 
 
146
  ## License
147
- - The license for the original implementation of Posenet-Mobilenet-Quantized can be found
148
- [here](https://github.com/rwightman/posenet-pytorch/blob/master/LICENSE.txt).
149
- - 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)
 
150
 
151
  ## References
152
  * [PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model](https://arxiv.org/abs/1803.08225)
153
  * [Source Model Implementation](https://github.com/rwightman/posenet-pytorch)
154
 
 
 
155
  ## Community
156
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
157
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
 
17
 
18
  Posenet performs pose estimation on human images.
19
 
20
+ This model is an implementation of Posenet-Mobilenet-Quantized found [here]({source_repo}).
21
  This repository provides scripts to run Posenet-Mobilenet-Quantized on Qualcomm® devices.
22
  More details on model performance across various devices, can be found
23
  [here](https://aihub.qualcomm.com/models/posenet_mobilenet_quantized).
 
32
  - Number of parameters: 3.31M
33
  - Model size: 3.47 MB
34
 
35
+ | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
36
+ |---|---|---|---|---|---|---|---|---|
37
+ | Posenet-Mobilenet-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 0.558 ms | 0 - 2 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
38
+ | Posenet-Mobilenet-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 0.64 ms | 0 - 11 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.so](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.so) |
39
+ | Posenet-Mobilenet-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 0.48 ms | 0 - 47 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
40
+ | Posenet-Mobilenet-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 0.445 ms | 0 - 18 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.so](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.so) |
41
+ | Posenet-Mobilenet-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 2.182 ms | 0 - 27 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
42
+ | Posenet-Mobilenet-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | QNN | 2.902 ms | 0 - 8 MB | INT8 | NPU | Use Export Script |
43
+ | Posenet-Mobilenet-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE | 12.597 ms | 0 - 12 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
44
+ | Posenet-Mobilenet-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 0.551 ms | 0 - 1 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
45
+ | Posenet-Mobilenet-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 0.555 ms | 0 - 2 MB | INT8 | NPU | Use Export Script |
46
+ | Posenet-Mobilenet-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 0.56 ms | 0 - 106 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
47
+ | Posenet-Mobilenet-Quantized | SA8255 (Proxy) | SA8255P Proxy | QNN | 0.561 ms | 0 - 2 MB | INT8 | NPU | Use Export Script |
48
+ | Posenet-Mobilenet-Quantized | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 0.557 ms | 0 - 17 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
49
+ | Posenet-Mobilenet-Quantized | SA8775 (Proxy) | SA8775P Proxy | QNN | 0.56 ms | 0 - 2 MB | INT8 | NPU | Use Export Script |
50
+ | Posenet-Mobilenet-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 0.559 ms | 0 - 3 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
51
+ | Posenet-Mobilenet-Quantized | SA8650 (Proxy) | SA8650P Proxy | QNN | 0.561 ms | 0 - 2 MB | INT8 | NPU | Use Export Script |
52
+ | Posenet-Mobilenet-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 0.714 ms | 0 - 50 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
53
+ | Posenet-Mobilenet-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 0.794 ms | 0 - 22 MB | INT8 | NPU | Use Export Script |
54
+ | Posenet-Mobilenet-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 0.412 ms | 0 - 26 MB | INT8 | NPU | [Posenet-Mobilenet-Quantized.tflite](https://huggingface.co/qualcomm/Posenet-Mobilenet-Quantized/blob/main/Posenet-Mobilenet-Quantized.tflite) |
55
+ | Posenet-Mobilenet-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 0.484 ms | 0 - 17 MB | INT8 | NPU | Use Export Script |
56
+ | Posenet-Mobilenet-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 0.679 ms | 0 - 0 MB | INT8 | NPU | Use Export Script |
57
 
58
 
59
 
 
 
 
 
 
 
60
 
61
  ## Installation
62
 
 
111
  ```bash
112
  python -m qai_hub_models.models.posenet_mobilenet_quantized.export
113
  ```
 
114
  ```
115
+ Profiling Results
116
+ ------------------------------------------------------------
117
+ Posenet-Mobilenet-Quantized
118
+ Device : Samsung Galaxy S23 (13)
119
+ Runtime : TFLITE
120
+ Estimated inference time (ms) : 0.6
121
+ Estimated peak memory usage (MB): [0, 2]
122
+ Total # Ops : 48
123
+ Compute Unit(s) : NPU (48 ops)
124
  ```
125
 
126
 
 
159
  Get more details on Posenet-Mobilenet-Quantized's performance across various devices [here](https://aihub.qualcomm.com/models/posenet_mobilenet_quantized).
160
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
161
 
162
+
163
  ## License
164
+ * The license for the original implementation of Posenet-Mobilenet-Quantized can be found [here](https://github.com/rwightman/posenet-pytorch/blob/master/LICENSE.txt).
165
+ * 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)
166
+
167
+
168
 
169
  ## References
170
  * [PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model](https://arxiv.org/abs/1803.08225)
171
  * [Source Model Implementation](https://github.com/rwightman/posenet-pytorch)
172
 
173
+
174
+
175
  ## Community
176
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
177
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).