Hugging Face's logo

qualcomm
/
MediaPipe-Pose-Estimation

Image Classification
PyTorch
TF Lite
real_time
android
Model card Files Community
qaihm-bot commited on
Commit
c6f3f38
•
1 Parent(s): 57f5052

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +20 -21
README.md CHANGED
@@ -34,10 +34,10 @@ More details on model performance across various devices, can be found
34
 
35
  | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
36
  | ---|---|---|---|---|---|---|---|
37
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 0.807 ms | 0 - 2 MB | FP16 | NPU | [MediaPipePoseDetector.tflite](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseDetector.tflite)
38
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 1.023 ms | 0 - 3 MB | FP16 | NPU | [MediaPipePoseLandmarkDetector.tflite](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseLandmarkDetector.tflite)
39
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 0.865 ms | 0 - 63 MB | FP16 | NPU | [MediaPipePoseDetector.so](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseDetector.so)
40
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 1.101 ms | 0 - 142 MB | FP16 | NPU | [MediaPipePoseLandmarkDetector.so](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseLandmarkDetector.so)
41
 
42
 
43
  ## Installation
@@ -45,11 +45,10 @@ More details on model performance across various devices, can be found
45
  This model can be installed as a Python package via pip.
46
 
47
  ```bash
48
- pip install "qai-hub-models[mediapipe_pose]"
49
  ```
50
 
51
 
52
-
53
  ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
54
 
55
  Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
@@ -98,31 +97,31 @@ python -m qai_hub_models.models.mediapipe_pose.export
98
  ```
99
  Profile Job summary of MediaPipePoseDetector
100
  --------------------------------------------------
101
- Device: Samsung Galaxy S23 Ultra (13)
102
- Estimated Inference Time: 0.81 ms
103
- Estimated Peak Memory Range: 0.03-1.57 MB
104
  Compute Units: NPU (106) | Total (106)
105
 
106
  Profile Job summary of MediaPipePoseLandmarkDetector
107
  --------------------------------------------------
108
- Device: Samsung Galaxy S23 Ultra (13)
109
- Estimated Inference Time: 1.02 ms
110
- Estimated Peak Memory Range: 0.01-3.10 MB
111
  Compute Units: NPU (229) | Total (229)
112
 
113
  Profile Job summary of MediaPipePoseDetector
114
  --------------------------------------------------
115
- Device: Samsung Galaxy S23 Ultra (13)
116
- Estimated Inference Time: 0.86 ms
117
- Estimated Peak Memory Range: 0.20-63.21 MB
118
- Compute Units: NPU (139) | Total (139)
119
 
120
  Profile Job summary of MediaPipePoseLandmarkDetector
121
  --------------------------------------------------
122
- Device: Samsung Galaxy S23 Ultra (13)
123
- Estimated Inference Time: 1.10 ms
124
- Estimated Peak Memory Range: 0.02-142.47 MB
125
- Compute Units: NPU (305) | Total (305)
126
 
127
 
128
  ```
@@ -227,7 +226,7 @@ Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
227
  ## License
228
  - The license for the original implementation of MediaPipe-Pose-Estimation can be found
229
  [here](https://github.com/zmurez/MediaPipePyTorch/blob/master/LICENSE).
230
- - 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).
231
 
232
  ## References
233
  * [BlazePose: On-device Real-time Body Pose tracking](https://arxiv.org/abs/2006.10204)
 
34
 
35
  | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
36
  | ---|---|---|---|---|---|---|---|
37
+ | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 0.806 ms | 0 - 2 MB | FP16 | NPU | [MediaPipePoseDetector.tflite](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseDetector.tflite)
38
+ | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 1.052 ms | 0 - 3 MB | FP16 | NPU | [MediaPipePoseLandmarkDetector.tflite](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseLandmarkDetector.tflite)
39
+ | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 0.808 ms | 0 - 5 MB | FP16 | NPU | [MediaPipePoseDetector.so](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseDetector.so)
40
+ | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 1.063 ms | 0 - 3 MB | FP16 | NPU | [MediaPipePoseLandmarkDetector.so](https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation/blob/main/MediaPipePoseLandmarkDetector.so)
41
 
42
 
43
  ## Installation
 
45
  This model can be installed as a Python package via pip.
46
 
47
  ```bash
48
+ pip install qai-hub-models
49
  ```
50
 
51
 
 
52
  ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
53
 
54
  Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
 
97
  ```
98
  Profile Job summary of MediaPipePoseDetector
99
  --------------------------------------------------
100
+ Device: Samsung Galaxy S24 (14)
101
+ Estimated Inference Time: 0.58 ms
102
+ Estimated Peak Memory Range: 0.06-37.81 MB
103
  Compute Units: NPU (106) | Total (106)
104
 
105
  Profile Job summary of MediaPipePoseLandmarkDetector
106
  --------------------------------------------------
107
+ Device: Samsung Galaxy S24 (14)
108
+ Estimated Inference Time: 0.76 ms
109
+ Estimated Peak Memory Range: 0.01-80.71 MB
110
  Compute Units: NPU (229) | Total (229)
111
 
112
  Profile Job summary of MediaPipePoseDetector
113
  --------------------------------------------------
114
+ Device: Samsung Galaxy S24 (14)
115
+ Estimated Inference Time: 0.58 ms
116
+ Estimated Peak Memory Range: 0.06-38.15 MB
117
+ Compute Units: NPU (106) | Total (106)
118
 
119
  Profile Job summary of MediaPipePoseLandmarkDetector
120
  --------------------------------------------------
121
+ Device: Samsung Galaxy S24 (14)
122
+ Estimated Inference Time: 0.77 ms
123
+ Estimated Peak Memory Range: 0.01-80.47 MB
124
+ Compute Units: NPU (229) | Total (229)
125
 
126
 
127
  ```
 
226
  ## License
227
  - The license for the original implementation of MediaPipe-Pose-Estimation can be found
228
  [here](https://github.com/zmurez/MediaPipePyTorch/blob/master/LICENSE).
229
+ - The license for the compiled assets for on-device deployment can be found [here]({deploy_license_url})
230
 
231
  ## References
232
  * [BlazePose: On-device Real-time Body Pose tracking](https://arxiv.org/abs/2006.10204)