Hugging Face's logo

qualcomm
/
Video-MAE

Video Classification
PyTorch
backbone
android
Model card Files
xet
 Community
qaihm-bot commited on
Commit
cbcfae9
·
verified ·
1 Parent(s): f397dbf

v0.46.1

Browse files

See https://github.com/quic/ai-hub-models/releases/v0.46.1 for changelog.

Files changed (5) hide show
  1. README.md +75 -212
  2. Video-MAE_float.dlc +0 -3
  3. Video-MAE_float.onnx.zip +0 -3
  4. Video-MAE_float.tflite +0 -3
  5. tool-versions.yaml +0 -4
README.md CHANGED
@@ -10,230 +10,93 @@ pipeline_tag: video-classification
10
 
11
  ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/web-assets/model_demo.png)
12
 
13
- # Video-MAE: Optimized for Mobile Deployment
14
- ## Sports and human action recognition in videos
15
-
16
 
17
  Video MAE (Masked Auto Encoder) is a network for doing video classification that uses the ViT (Vision Transformer) backbone.
18
 
19
- This model is an implementation of Video-MAE found [here](https://github.com/MCG-NJU/VideoMAE).
20
-
21
-
22
- This repository provides scripts to run Video-MAE on Qualcomm® devices.
23
- More details on model performance across various devices, can be found
24
- [here](https://aihub.qualcomm.com/models/video_mae).
25
-
26
-
27
-
28
- ### Model Details
29
-
30
- - **Model Type:** Model_use_case.video_classification
31
- - **Model Stats:**
32
- - Model checkpoint: Kinectics-400
33
- - Input resolution: 224x224
34
- - Number of parameters: 87.7M
35
- - Model size (float): 335 MB
36
-
37
- | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
38
- |---|---|---|---|---|---|---|---|---|
39
- | Video-MAE | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 423.568 ms | 0 - 908 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
40
- | Video-MAE | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 1136.783 ms | 4 - 851 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
41
- | Video-MAE | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 291.792 ms | 1 - 1054 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
42
- | Video-MAE | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 575.304 ms | 9 - 1005 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
43
- | Video-MAE | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 139.117 ms | 0 - 5 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
44
- | Video-MAE | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 495.828 ms | 9 - 12 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
45
- | Video-MAE | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 572.575 ms | 0 - 217 MB | NPU | [Video-MAE.onnx.zip](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.onnx.zip) |
46
- | Video-MAE | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 723.049 ms | 0 - 901 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
47
- | Video-MAE | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 501.186 ms | 2 - 887 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
48
- | Video-MAE | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 423.568 ms | 0 - 908 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
49
- | Video-MAE | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 1136.783 ms | 4 - 851 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
50
- | Video-MAE | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 669.5 ms | 0 - 845 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
51
- | Video-MAE | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 723.049 ms | 0 - 901 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
52
- | Video-MAE | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 501.186 ms | 2 - 887 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
53
- | Video-MAE | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 100.899 ms | 1 - 1109 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
54
- | Video-MAE | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 349.189 ms | 9 - 1059 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
55
- | Video-MAE | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 416.581 ms | 1 - 1183 MB | NPU | [Video-MAE.onnx.zip](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.onnx.zip) |
56
- | Video-MAE | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 74.924 ms | 0 - 899 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
57
- | Video-MAE | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 259.329 ms | 9 - 867 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
58
- | Video-MAE | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 439.219 ms | 1 - 966 MB | NPU | [Video-MAE.onnx.zip](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.onnx.zip) |
59
- | Video-MAE | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | TFLITE | 59.82 ms | 1 - 913 MB | NPU | [Video-MAE.tflite](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.tflite) |
60
- | Video-MAE | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | QNN_DLC | 294.871 ms | 11 - 902 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
61
- | Video-MAE | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | ONNX | 593.719 ms | 9 - 989 MB | NPU | [Video-MAE.onnx.zip](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.onnx.zip) |
62
- | Video-MAE | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 516.587 ms | 9 - 9 MB | NPU | [Video-MAE.dlc](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.dlc) |
63
- | Video-MAE | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 599.153 ms | 187 - 187 MB | NPU | [Video-MAE.onnx.zip](https://huggingface.co/qualcomm/Video-MAE/blob/main/Video-MAE.onnx.zip) |
64
-
65
-
66
-
67
-
68
- ## Installation
69
-
70
-
71
- Install the package via pip:
72
- ```bash
73
- # NOTE: 3.10 <= PYTHON_VERSION < 3.14 is supported.
74
- pip install "qai-hub-models[video-mae]"
75
- ```
76
-
77
-
78
- ## Configure Qualcomm® AI Hub Workbench to run this model on a cloud-hosted device
79
-
80
- Sign-in to [Qualcomm® AI Hub Workbench](https://workbench.aihub.qualcomm.com/) with your
81
- Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
82
-
83
- With this API token, you can configure your client to run models on the cloud
84
- hosted devices.
85
- ```bash
86
- qai-hub configure --api_token API_TOKEN
87
- ```
88
- Navigate to [docs](https://workbench.aihub.qualcomm.com/docs/) for more information.
89
-
90
-
91
-
92
- ## Demo off target
93
-
94
- The package contains a simple end-to-end demo that downloads pre-trained
95
- weights and runs this model on a sample input.
96
-
97
- ```bash
98
- python -m qai_hub_models.models.video_mae.demo
99
- ```
100
-
101
- The above demo runs a reference implementation of pre-processing, model
102
- inference, and post processing.
103
-
104
- **NOTE**: If you want running in a Jupyter Notebook or Google Colab like
105
- environment, please add the following to your cell (instead of the above).
106
- ```
107
- %run -m qai_hub_models.models.video_mae.demo
108
- ```
109
-
110
-
111
- ### Run model on a cloud-hosted device
112
-
113
- In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
114
- device. This script does the following:
115
- * Performance check on-device on a cloud-hosted device
116
- * Downloads compiled assets that can be deployed on-device for Android.
117
- * Accuracy check between PyTorch and on-device outputs.
118
-
119
- ```bash
120
- python -m qai_hub_models.models.video_mae.export
121
- ```
122
-
123
-
124
-
125
- ## How does this work?
126
-
127
- This [export script](https://aihub.qualcomm.com/models/video_mae/qai_hub_models/models/Video-MAE/export.py)
128
- leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
129
- on-device. Lets go through each step below in detail:
130
-
131
- Step 1: **Compile model for on-device deployment**
132
-
133
- To compile a PyTorch model for on-device deployment, we first trace the model
134
- in memory using the `jit.trace` and then call the `submit_compile_job` API.
135
-
136
- ```python
137
- import torch
138
-
139
- import qai_hub as hub
140
- from qai_hub_models.models.video_mae import Model
141
-
142
- # Load the model
143
- torch_model = Model.from_pretrained()
144
-
145
- # Device
146
- device = hub.Device("Samsung Galaxy S25")
147
-
148
- # Trace model
149
- input_shape = torch_model.get_input_spec()
150
- sample_inputs = torch_model.sample_inputs()
151
-
152
- pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
153
-
154
- # Compile model on a specific device
155
- compile_job = hub.submit_compile_job(
156
- model=pt_model,
157
- device=device,
158
- input_specs=torch_model.get_input_spec(),
159
- )
160
-
161
- # Get target model to run on-device
162
- target_model = compile_job.get_target_model()
163
-
164
- ```
165
-
166
-
167
- Step 2: **Performance profiling on cloud-hosted device**
168
-
169
- After compiling models from step 1. Models can be profiled model on-device using the
170
- `target_model`. Note that this scripts runs the model on a device automatically
171
- provisioned in the cloud. Once the job is submitted, you can navigate to a
172
- provided job URL to view a variety of on-device performance metrics.
173
- ```python
174
- profile_job = hub.submit_profile_job(
175
- model=target_model,
176
- device=device,
177
- )
178
-
179
- ```
180
-
181
- Step 3: **Verify on-device accuracy**
182
-
183
- To verify the accuracy of the model on-device, you can run on-device inference
184
- on sample input data on the same cloud hosted device.
185
- ```python
186
- input_data = torch_model.sample_inputs()
187
- inference_job = hub.submit_inference_job(
188
- model=target_model,
189
- device=device,
190
- inputs=input_data,
191
- )
192
- on_device_output = inference_job.download_output_data()
193
-
194
- ```
195
- With the output of the model, you can compute like PSNR, relative errors or
196
- spot check the output with expected output.
197
-
198
- **Note**: This on-device profiling and inference requires access to Qualcomm®
199
- AI Hub Workbench. [Sign up for access](https://myaccount.qualcomm.com/signup).
200
-
201
-
202
-
203
-
204
- ## Deploying compiled model to Android
205
-
206
-
207
- The models can be deployed using multiple runtimes:
208
- - TensorFlow Lite (`.tflite` export): [This
209
- tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
210
- guide to deploy the .tflite model in an Android application.
211
-
212
-
213
- - QNN (`.so` export ): This [sample
214
- app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
215
- provides instructions on how to use the `.so` shared library in an Android application.
216
-
217
-
218
- ## View on Qualcomm® AI Hub
219
- Get more details on Video-MAE's performance across various devices [here](https://aihub.qualcomm.com/models/video_mae).
220
- Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
221
-
222
 
223
  ## License
224
  * The license for the original implementation of Video-MAE can be found
225
  [here](https://github.com/MCG-NJU/VideoMAE/blob/main/LICENSE).
226
 
227
-
228
-
229
  ## References
230
  * [Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602)
231
  * [Source Model Implementation](https://github.com/MCG-NJU/VideoMAE)
232
 
233
-
234
-
235
  ## Community
236
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
237
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
238
-
239
-
 
10
 
11
  ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/web-assets/model_demo.png)
12
 
13
+ # Video-MAE: Optimized for Qualcomm Devices
 
 
14
 
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/quic/ai-hub-models/blob/main/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
+
22
+ ## Getting Started
23
+ There are two ways to deploy this model on your device:
24
+
25
+ ### Option 1: Download Pre-Exported Models
26
+
27
+ Below are pre-exported model assets ready for deployment.
28
+
29
+ | Runtime | Precision | Chipset | SDK Versions | Download |
30
+ |---|---|---|---|---|
31
+ | ONNX | float | Universal | QAIRT 2.37, ONNX Runtime 1.23.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.46.1/video_mae-onnx-float.zip)
32
+ | ONNX | w8a16 | Universal | QAIRT 2.37, ONNX Runtime 1.23.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.46.1/video_mae-onnx-w8a16.zip)
33
+ | QNN_DLC | float | Universal | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.46.1/video_mae-qnn_dlc-float.zip)
34
+ | QNN_DLC | w8a16 | Universal | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.46.1/video_mae-qnn_dlc-w8a16.zip)
35
+ | TFLITE | float | Universal | QAIRT 2.42, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/video_mae/releases/v0.46.1/video_mae-tflite-float.zip)
36
+
37
+ For more device-specific assets and performance metrics, visit **[Video-MAE on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/video_mae)**.
38
+
39
+
40
+ ### Option 2: Export with Custom Configurations
41
+
42
+ Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/video_mae) Python library to compile and export the model with your own:
43
+ - Custom weights (e.g., fine-tuned checkpoints)
44
+ - Custom input shapes
45
+ - Target device and runtime configurations
46
+
47
+ This option is ideal if you need to customize the model beyond the default configuration provided here.
48
+
49
+ See our repository for [Video-MAE on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/video_mae) for usage instructions.
50
+
51
+ ## Model Details
52
+
53
+ **Model Type:** Model_use_case.video_classification
54
+
55
+ **Model Stats:**
56
+ - Model checkpoint: Kinectics-400
57
+ - Input resolution: 224x224
58
+ - Number of parameters: 87.7M
59
+ - Model size (float): 335 MB
60
+
61
+ ## Performance Summary
62
+ | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
63
+ |---|---|---|---|---|---|---
64
+ | Video-MAE | ONNX | float | Snapdragon® X Elite | 599.027 ms | 187 - 187 MB | NPU
65
+ | Video-MAE | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 417.051 ms | 10 - 1193 MB | NPU
66
+ | Video-MAE | ONNX | float | Qualcomm® QCS8550 (Proxy) | 572.814 ms | 0 - 217 MB | NPU
67
+ | Video-MAE | ONNX | float | Qualcomm® QCS9075 | 672.39 ms | 9 - 21 MB | NPU
68
+ | Video-MAE | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 439.593 ms | 1 - 971 MB | NPU
69
+ | Video-MAE | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 594.011 ms | 1 - 977 MB | NPU
70
+ | Video-MAE | QNN_DLC | float | Snapdragon® X Elite | 473.072 ms | 9 - 9 MB | NPU
71
+ | Video-MAE | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 354.537 ms | 9 - 1124 MB | NPU
72
+ | Video-MAE | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 1038.798 ms | 0 - 914 MB | NPU
73
+ | Video-MAE | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 454.586 ms | 9 - 12 MB | NPU
74
+ | Video-MAE | QNN_DLC | float | Qualcomm® SA8775P | 485.216 ms | 0 - 956 MB | NPU
75
+ | Video-MAE | QNN_DLC | float | Qualcomm® QCS9075 | 515.489 ms | 11 - 22 MB | NPU
76
+ | Video-MAE | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 579.173 ms | 9 - 1066 MB | NPU
77
+ | Video-MAE | QNN_DLC | float | Qualcomm® SA7255P | 1038.798 ms | 0 - 914 MB | NPU
78
+ | Video-MAE | QNN_DLC | float | Qualcomm® SA8295P | 565.13 ms | 0 - 858 MB | NPU
79
+ | Video-MAE | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 331.251 ms | 9 - 928 MB | NPU
80
+ | Video-MAE | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 249.902 ms | 11 - 966 MB | NPU
81
+ | Video-MAE | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 100.537 ms | 0 - 1169 MB | NPU
82
+ | Video-MAE | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 5536.906 ms | 42 - 59 MB | CPU
83
+ | Video-MAE | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 138.298 ms | 0 - 4 MB | NPU
84
+ | Video-MAE | TFLITE | float | Qualcomm® SA8775P | 160.51 ms | 0 - 961 MB | NPU
85
+ | Video-MAE | TFLITE | float | Qualcomm® QCS9075 | 172.102 ms | 0 - 207 MB | NPU
86
+ | Video-MAE | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 298.251 ms | 0 - 1115 MB | NPU
87
+ | Video-MAE | TFLITE | float | Qualcomm® SA7255P | 5536.906 ms | 42 - 59 MB | CPU
88
+ | Video-MAE | TFLITE | float | Qualcomm® SA8295P | 214.16 ms | 0 - 912 MB | NPU
89
+ | Video-MAE | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 74.663 ms | 0 - 962 MB | NPU
90
+ | Video-MAE | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 60.822 ms | 0 - 954 MB | NPU
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
91
 
92
  ## License
93
  * The license for the original implementation of Video-MAE can be found
94
  [here](https://github.com/MCG-NJU/VideoMAE/blob/main/LICENSE).
95
 
 
 
96
  ## References
97
  * [Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602)
98
  * [Source Model Implementation](https://github.com/MCG-NJU/VideoMAE)
99
 
 
 
100
  ## Community
101
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
102
  * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
 
 
Video-MAE_float.dlc DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:5809fa48eda668973b39cd9d074507cb4f3142ce11a9bf31e4ee5e3c96a8df49
3
- size 351314884
 
 
 
 
Video-MAE_float.onnx.zip DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c42437267f75e065a1c448a9e1a4ee0759b1e5634aba9c1fdfb80a9b06d014f7
3
- size 221427620
 
 
 
 
Video-MAE_float.tflite DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:05e2c7e3ae7a1ccc588bf07bcfa47af934ed473b9259558339d23a394fe7d2b3
3
- size 351051860
 
 
 
 
tool-versions.yaml DELETED
@@ -1,4 +0,0 @@
1
- tool_versions:
2
- onnx:
3
- qairt: 2.37.1.250807093845_124904
4
- onnx_runtime: 1.23.0