Hugging Face's logo

qualcomm
/
Video-MAE

Video Classification
PyTorch
backbone
android
Model card Files
xet
 Community
qaihm-bot commited on
Commit
d0e533d
·
verified ·
1 Parent(s): 31d5125

v0.58.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.58.0 for changelog.

Files changed (2) hide show
  1. README.md +36 -32
  2. release_assets.json +4 -4
README.md CHANGED
@@ -15,7 +15,7 @@ pipeline_tag: video-classification
15
  Video MAE (Masked Auto Encoder) is a network for doing video classification that uses the ViT (Vision Transformer) backbone.
16
 
17
  This is based on the implementation of Video-MAE found [here](https://github.com/MCG-NJU/VideoMAE).
18
- This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.57.3/src/qai_hub_models/models/video_mae) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
19
 
20
  Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
21
 
@@ -28,23 +28,23 @@ Below are pre-exported model assets ready for deployment.
28
 
29
  | Runtime | Precision | Chipset | SDK Versions | Download |
30
  |---|---|---|---|---|
31
- | ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.57.3/video_mae-onnx-float.zip)
32
- | QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.57.3/video_mae-qnn_dlc-float.zip)
33
- | TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.57.3/video_mae-tflite-float.zip)
34
 
35
  For more device-specific assets and performance metrics, visit **[Video-MAE on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/video_mae)**.
36
 
37
 
38
  ### Option 2: Export with Custom Configurations
39
 
40
- Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.57.3/src/qai_hub_models/models/video_mae) Python library to compile and export the model with your own:
41
  - Custom weights (e.g., fine-tuned checkpoints)
42
  - Custom input shapes
43
  - Target device and runtime configurations
44
 
45
  This option is ideal if you need to customize the model beyond the default configuration provided here.
46
 
47
- See our repository for [Video-MAE on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.57.3/src/qai_hub_models/models/video_mae) for usage instructions.
48
 
49
  ## Model Details
50
 
@@ -59,32 +59,36 @@ See our repository for [Video-MAE on GitHub](https://github.com/qualcomm/ai-hub-
59
  ## Performance Summary
60
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
61
  |---|---|---|---|---|---|---
62
- | Video-MAE | ONNX | float | Snapdragon® X2 Elite | 1573.648 ms | 167 - 167 MB | NPU
63
- | Video-MAE | ONNX | float | Snapdragon® X Elite | 2876.574 ms | 193 - 193 MB | NPU
64
- | Video-MAE | ONNX | float | Qualcomm® QCS8550 (Proxy) | 2818.888 ms | 1 - 213 MB | NPU
65
- | Video-MAE | ONNX | float | Snapdragon® 8 Elite Mobile | 1649.301 ms | 4 - 5496 MB | NPU
66
- | Video-MAE | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1516.515 ms | 1 - 5696 MB | NPU
67
- | Video-MAE | ONNX | float | Qualcomm® QCS9075 | 5020.553 ms | 46 - 137 MB | NPU
68
- | Video-MAE | ONNX | float | Qualcomm® QCS8750 | 1649.301 ms | 4 - 5496 MB | NPU
69
- | Video-MAE | ONNX | float | Qualcomm® QCS7181 | 2876.574 ms | 193 - 193 MB | NPU
70
- | Video-MAE | QNN_DLC | float | Snapdragon® X2 Elite | 1968.594 ms | 46 - 46 MB | NPU
71
- | Video-MAE | QNN_DLC | float | Snapdragon® X Elite | 3275.201 ms | 46 - 46 MB | NPU
72
- | Video-MAE | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 5135.131 ms | 46 - 50 MB | NPU
73
- | Video-MAE | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 3552.984 ms | 1 - 6093 MB | NPU
74
- | Video-MAE | QNN_DLC | float | Qualcomm® SA8295P | 3812.661 ms | 36 - 5739 MB | NPU
75
- | Video-MAE | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 4070.911 ms | 3 - 6272 MB | NPU
76
- | Video-MAE | QNN_DLC | float | Qualcomm® QCS9075 | 5458.899 ms | 46 - 94 MB | NPU
77
- | Video-MAE | QNN_DLC | float | Qualcomm® QCS8750 | 3552.984 ms | 1 - 6093 MB | NPU
78
- | Video-MAE | QNN_DLC | float | Qualcomm® QCS7181 | 3275.201 ms | 46 - 46 MB | NPU
79
- | Video-MAE | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 5157.258 ms | 1 - 5 MB | NPU
80
- | Video-MAE | TFLITE | float | Qualcomm® SA8775P | 14234.218 ms | 188 - 197 MB | CPU
81
- | Video-MAE | TFLITE | float | Qualcomm® SA8650P | 14234.218 ms | 188 - 197 MB | CPU
82
- | Video-MAE | TFLITE | float | Qualcomm® SA8255P | 14234.218 ms | 188 - 197 MB | CPU
83
- | Video-MAE | TFLITE | float | Snapdragon® 8 Elite Mobile | 3536.957 ms | 1 - 5953 MB | NPU
84
- | Video-MAE | TFLITE | float | Qualcomm® SA8295P | 3741.683 ms | 1 - 5603 MB | NPU
85
- | Video-MAE | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 4086.474 ms | 1 - 6105 MB | NPU
86
- | Video-MAE | TFLITE | float | Qualcomm® QCS9075 | 5389.455 ms | 0 - 280 MB | NPU
87
- | Video-MAE | TFLITE | float | Qualcomm® QCS8750 | 3536.957 ms | 1 - 5953 MB | NPU
 
 
 
 
88
 
89
  ## License
90
  * The license for the original implementation of Video-MAE can be found
 
15
  Video MAE (Masked Auto Encoder) is a network for doing video classification that uses the ViT (Vision Transformer) backbone.
16
 
17
  This is based on the implementation of Video-MAE found [here](https://github.com/MCG-NJU/VideoMAE).
18
+ This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/video_mae) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
19
 
20
  Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
21
 
 
28
 
29
  | Runtime | Precision | Chipset | SDK Versions | Download |
30
  |---|---|---|---|---|
31
+ | ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.58.0/video_mae-onnx-float.zip)
32
+ | QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.58.0/video_mae-qnn_dlc-float.zip)
33
+ | TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.58.0/video_mae-tflite-float.zip)
34
 
35
  For more device-specific assets and performance metrics, visit **[Video-MAE on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/video_mae)**.
36
 
37
 
38
  ### Option 2: Export with Custom Configurations
39
 
40
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/video_mae) Python library to compile and export the model with your own:
41
  - Custom weights (e.g., fine-tuned checkpoints)
42
  - Custom input shapes
43
  - Target device and runtime configurations
44
 
45
  This option is ideal if you need to customize the model beyond the default configuration provided here.
46
 
47
+ See our repository for [Video-MAE on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/video_mae) for usage instructions.
48
 
49
  ## Model Details
50
 
 
59
  ## Performance Summary
60
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
61
  |---|---|---|---|---|---|---
62
+ | Video-MAE | ONNX | float | Snapdragon® X2 Elite | 1579.018 ms | 46 - 46 MB | NPU
63
+ | Video-MAE | ONNX | float | Snapdragon® X Elite | 2876.701 ms | 192 - 192 MB | NPU
64
+ | Video-MAE | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2815.561 ms | 47 - 51 MB | NPU
65
+ | Video-MAE | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 6805.927 ms | 46 - 95 MB | NPU
66
+ | Video-MAE | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1515.987 ms | 0 - 5708 MB | NPU
67
+ | Video-MAE | ONNX | float | Snapdragon® 8 Elite Mobile | 1645.755 ms | 1 - 5502 MB | NPU
68
+ | Video-MAE | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 1645.755 ms | 1 - 5502 MB | NPU
69
+ | Video-MAE | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 2876.701 ms | 192 - 192 MB | NPU
70
+ | Video-MAE | QNN_DLC | float | Snapdragon® X2 Elite | 1967.699 ms | 46 - 46 MB | NPU
71
+ | Video-MAE | QNN_DLC | float | Snapdragon® X Elite | 3234.8 ms | 46 - 46 MB | NPU
72
+ | Video-MAE | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 5140.112 ms | 46 - 49 MB | NPU
73
+ | Video-MAE | QNN_DLC | float | Qualcomm® SA8775P | 5351.378 ms | 38 - 6133 MB | NPU
74
+ | Video-MAE | QNN_DLC | float | Qualcomm® SA8650P | 5351.378 ms | 38 - 6133 MB | NPU
75
+ | Video-MAE | QNN_DLC | float | Qualcomm® SA8255P | 5351.378 ms | 38 - 6133 MB | NPU
76
+ | Video-MAE | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 5288.535 ms | 46 - 94 MB | NPU
77
+ | Video-MAE | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 4069.479 ms | 8 - 6276 MB | NPU
78
+ | Video-MAE | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 3552.571 ms | 147 - 6236 MB | NPU
79
+ | Video-MAE | QNN_DLC | float | Qualcomm® SA8295P | 3815.309 ms | 46 - 5747 MB | NPU
80
+ | Video-MAE | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 3552.571 ms | 147 - 6236 MB | NPU
81
+ | Video-MAE | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 3234.8 ms | 46 - 46 MB | NPU
82
+ | Video-MAE | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 3512.692 ms | 1 - 6669 MB | NPU
83
+ | Video-MAE | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 5143.197 ms | 1 - 4 MB | NPU
84
+ | Video-MAE | TFLITE | float | Qualcomm® SA8775P | 5320.509 ms | 2 - 5971 MB | NPU
85
+ | Video-MAE | TFLITE | float | Qualcomm® SA8650P | 5320.509 ms | 2 - 5971 MB | NPU
86
+ | Video-MAE | TFLITE | float | Qualcomm® SA8255P | 5320.509 ms | 2 - 5971 MB | NPU
87
+ | Video-MAE | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 5258.82 ms | 0 - 279 MB | NPU
88
+ | Video-MAE | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 4088.26 ms | 1 - 6101 MB | NPU
89
+ | Video-MAE | TFLITE | float | Snapdragon® 8 Elite Mobile | 3540.576 ms | 1 - 5951 MB | NPU
90
+ | Video-MAE | TFLITE | float | Qualcomm® SA8295P | 3736.9 ms | 1 - 5604 MB | NPU
91
+ | Video-MAE | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 3540.576 ms | 1 - 5951 MB | NPU
92
 
93
  ## License
94
  * The license for the original implementation of Video-MAE can be found
release_assets.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "version": "0.57.3",
3
  "precisions": {
4
  "float": {
5
  "universal_assets": {
@@ -8,20 +8,20 @@
8
  "qairt": "2.45.0.260326154327",
9
  "litert": "1.4.4"
10
  },
11
- "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.57.3/video_mae-tflite-float.zip"
12
  },
13
  "qnn_dlc": {
14
  "tool_versions": {
15
  "qairt": "2.45.0.260326154327"
16
  },
17
- "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.57.3/video_mae-qnn_dlc-float.zip"
18
  },
19
  "onnx": {
20
  "tool_versions": {
21
  "qairt": "2.45.0.260326154327",
22
  "onnx_runtime": "1.25.0"
23
  },
24
- "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.57.3/video_mae-onnx-float.zip"
25
  }
26
  }
27
  }
 
1
  {
2
+ "version": "0.58.0",
3
  "precisions": {
4
  "float": {
5
  "universal_assets": {
 
8
  "qairt": "2.45.0.260326154327",
9
  "litert": "1.4.4"
10
  },
11
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.58.0/video_mae-tflite-float.zip"
12
  },
13
  "qnn_dlc": {
14
  "tool_versions": {
15
  "qairt": "2.45.0.260326154327"
16
  },
17
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.58.0/video_mae-qnn_dlc-float.zip"
18
  },
19
  "onnx": {
20
  "tool_versions": {
21
  "qairt": "2.45.0.260326154327",
22
  "onnx_runtime": "1.25.0"
23
  },
24
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.58.0/video_mae-onnx-float.zip"
25
  }
26
  }
27
  }