Hugging Face's logo

qualcomm
/
Movenet

Keypoint Detection
PyTorch
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
58edb71
·
verified ·
1 Parent(s): e74230f

v0.34.0

Browse files

See https://github.com/quic/ai-hub-models/releases/v0.34.0 for changelog.

Files changed (7) hide show
  1. Movenet.dlc +2 -2
  2. Movenet.onnx.zip +2 -2
  3. Movenet.tflite +2 -2
  4. README.md +31 -40
  5. precompiled/qualcomm-snapdragon-x-elite/Movenet.bin +2 -2
  6. precompiled/qualcomm-snapdragon-x-elite/Movenet.onnx.zip +2 -2
  7. precompiled/qualcomm-snapdragon-x-elite/sdk_versions.yml +5 -0
Movenet.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3b2d9413285dd9d8ad99e0c58e845eb59cd07f48c44dc625ef4a4c2c9c35151a
3
- size 9508617
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6ff452f2a1a6e9e32b1f5ccd8b032e3edd6784ba5b38b84f1a63da2411f69b9
3
+ size 9509181
Movenet.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e43d19004d14c082282b06ce9077d10cf8006b4af0ee003bfcd06e55cdfc187d
3
- size 8676352
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a69bfe31d2e68ab9b774988094dde19fda04485ec146455386ee044f196b162b
3
+ size 8676402
Movenet.tflite CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:38ca45ef17452cd0b4d2cb6b0d8fa2eb5b5bd1f33b52322e1a5ce20825201c02
3
- size 9343844
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5673dc872fdca8ef70ad0b2a811799144d9d2489289ec42b4901e598edf7359b
3
+ size 9344068
README.md CHANGED
@@ -23,6 +23,7 @@ More details on model performance across various devices, can be found
23
  [here](https://aihub.qualcomm.com/models/movenet).
24
 
25
 
 
26
  ### Model Details
27
 
28
  - **Model Type:** Model_use_case.pose_estimation
@@ -34,35 +35,35 @@ More details on model performance across various devices, can be found
34
 
35
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
36
  |---|---|---|---|---|---|---|---|---|
37
- | Movenet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 6.186 ms | 1 - 10 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
38
- | Movenet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 18.698 ms | 13 - 23 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
39
- | Movenet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 4.256 ms | 1 - 23 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
40
- | Movenet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 25.635 ms | 13 - 37 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
41
- | Movenet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 3.734 ms | 1 - 3 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
42
- | Movenet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 14.22 ms | 12 - 17 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
43
- | Movenet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 5.487 ms | 1 - 11 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
44
- | Movenet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 13.894 ms | 12 - 25 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
45
- | Movenet | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 6.186 ms | 1 - 10 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
46
- | Movenet | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 18.698 ms | 13 - 23 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
47
- | Movenet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 4.544 ms | 1 - 3 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
48
- | Movenet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 35.531 ms | 13 - 16 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
49
- | Movenet | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 3.873 ms | 0 - 17 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
50
- | Movenet | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 10.982 ms | 13 - 31 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
51
- | Movenet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 3.96 ms | 0 - 2 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
52
- | Movenet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 11.997 ms | 10 - 14 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
53
- | Movenet | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 5.487 ms | 1 - 11 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
54
- | Movenet | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 13.894 ms | 12 - 25 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
55
- | Movenet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 4.134 ms | 0 - 2 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
56
- | Movenet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 26.169 ms | 15 - 19 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
57
- | Movenet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 13.314 ms | 12 - 14 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
58
- | Movenet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 3.41 ms | 0 - 18 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
59
- | Movenet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 5.945 ms | 13 - 33 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
60
- | Movenet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 9.12 ms | 12 - 29 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
61
- | Movenet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 2.86 ms | 0 - 14 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
62
- | Movenet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 8.196 ms | 18 - 34 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
63
- | Movenet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 9.163 ms | 19 - 33 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
64
- | Movenet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 7.053 ms | 4 - 4 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
65
- | Movenet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 9.47 ms | 16 - 16 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
66
 
67
 
68
 
@@ -120,17 +121,7 @@ device. This script does the following:
120
  ```bash
121
  python -m qai_hub_models.models.movenet.export
122
  ```
123
- ```
124
- Profiling Results
125
- ------------------------------------------------------------
126
- Movenet
127
- Device : cs_8275 (ANDROID 14)
128
- Runtime : TFLITE
129
- Estimated inference time (ms) : 6.2
130
- Estimated peak memory usage (MB): [1, 10]
131
- Total # Ops : 152
132
- Compute Unit(s) : npu (0 ops) gpu (0 ops) cpu (152 ops)
133
- ```
134
 
135
 
136
  ## How does this work?
 
23
  [here](https://aihub.qualcomm.com/models/movenet).
24
 
25
 
26
+
27
  ### Model Details
28
 
29
  - **Model Type:** Model_use_case.pose_estimation
 
35
 
36
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
+ | Movenet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 6.231 ms | 1 - 10 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
39
+ | Movenet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 21.402 ms | 10 - 21 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
40
+ | Movenet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 5.915 ms | 1 - 24 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
41
+ | Movenet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 18.175 ms | 13 - 38 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
42
+ | Movenet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 3.776 ms | 0 - 3 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
43
+ | Movenet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 11.42 ms | 12 - 16 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
44
+ | Movenet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 5.507 ms | 1 - 13 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
45
+ | Movenet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 13.692 ms | 13 - 26 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
46
+ | Movenet | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 6.231 ms | 1 - 10 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
47
+ | Movenet | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 21.402 ms | 10 - 21 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
48
+ | Movenet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 3.743 ms | 0 - 2 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
49
+ | Movenet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 11.347 ms | 13 - 16 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
50
+ | Movenet | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 3.605 ms | 0 - 18 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
51
+ | Movenet | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 12.447 ms | 13 - 31 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
52
+ | Movenet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 3.783 ms | 0 - 10 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
53
+ | Movenet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 11.324 ms | 12 - 16 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
54
+ | Movenet | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 5.507 ms | 1 - 13 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
55
+ | Movenet | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 13.692 ms | 13 - 26 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
56
+ | Movenet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 3.782 ms | 0 - 3 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
57
+ | Movenet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 14.844 ms | 12 - 16 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
58
+ | Movenet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 10.622 ms | 11 - 14 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
59
+ | Movenet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 3.472 ms | 0 - 20 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
60
+ | Movenet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 5.567 ms | 13 - 34 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
61
+ | Movenet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 9.376 ms | 12 - 32 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
62
+ | Movenet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 2.963 ms | 7 - 21 MB | CPU | [Movenet.tflite](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.tflite) |
63
+ | Movenet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 8.569 ms | 18 - 34 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
64
+ | Movenet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 8.807 ms | 19 - 32 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
65
+ | Movenet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 6.632 ms | 4 - 4 MB | CPU | [Movenet.dlc](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.dlc) |
66
+ | Movenet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 8.657 ms | 16 - 16 MB | CPU | [Movenet.onnx](https://huggingface.co/qualcomm/Movenet/blob/main/Movenet.onnx) |
67
 
68
 
69
 
 
121
  ```bash
122
  python -m qai_hub_models.models.movenet.export
123
  ```
124
+
 
 
 
 
 
 
 
 
 
 
125
 
126
 
127
  ## How does this work?
precompiled/qualcomm-snapdragon-x-elite/Movenet.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:29e46a152da761088d4dce09df41bb6b2067a845260bd031564a138655fd88b3
3
- size 6191320
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29cf26fdeddfae0544e22d186eef64ff5620c9515daa6032a207180055f1c2d9
3
+ size 6240496
precompiled/qualcomm-snapdragon-x-elite/Movenet.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:174933be43ecb455cddc68b9856c458f30d22ba6fbb1a0667b0d69faa75041e2
3
- size 4467316
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:566ec4fb20f415bfcbc87e25f2f5ad7668f4f5b2bf683ba7acf19f0c8f760fa3
3
+ size 4476093
precompiled/qualcomm-snapdragon-x-elite/sdk_versions.yml ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ sdk_versions:
2
+ qnn_context_binary:
3
+ qairt: 2.34.2.250528164111_119506
4
+ precompiled_qnn_onnx:
5
+ qairt: 2.33.2.250410134701_117956