Hugging Face's logo

qualcomm
/
Mobile-VIT

Image Classification
PyTorch
backbone
android
Model card Files
xet
Community
Mobile-VIT
File size: 7,308 Bytes 
9511cdd
 
 
 
 
 
 
 
 
 
 
 
008f1af
9511cdd
 
 
008f1af
 
 
 
 
 
 
 
 
 
 
 
 
 
c397e71
 
 
 
 
008f1af
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9511cdd
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
 
---
library_name: pytorch
license: other
tags:
- backbone
- android
pipeline_tag: image-classification

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/mobile_vit/web-assets/model_demo.png)

# Mobile-VIT: Optimized for Qualcomm Devices

MobileVit is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.

This is based on the implementation of Mobile-VIT found [here](https://github.com/apple/ml-cvnets).
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/mobile_vit) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).

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.

## Getting Started
There are two ways to deploy this model on your device:

### Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

| Runtime | Precision | Chipset | SDK Versions | Download |
|---|---|---|---|---|
| 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/mobile_vit/releases/v0.46.0/mobile_vit-onnx-float.zip)
| 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/mobile_vit/releases/v0.46.0/mobile_vit-onnx-w8a16.zip)
| QNN_DLC | float | Universal | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/mobile_vit/releases/v0.46.0/mobile_vit-qnn_dlc-float.zip)
| QNN_DLC | w8a16 | Universal | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/mobile_vit/releases/v0.46.0/mobile_vit-qnn_dlc-w8a16.zip)
| 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/mobile_vit/releases/v0.46.0/mobile_vit-tflite-float.zip)

For more device-specific assets and performance metrics, visit **[Mobile-VIT on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/mobile_vit)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/mobile_vit) Python library to compile and export the model with your own:
- Custom weights (e.g., fine-tuned checkpoints)
- Custom input shapes
- Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for [Mobile-VIT on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/mobile_vit) for usage instructions.

## Model Details

**Model Type:** Model_use_case.image_classification

**Model Stats:**
- Model checkpoint: Imagenet
- Input resolution: 224x224
- Number of parameters: 5.57M
- Model size (float): 21.4 MB
- Model size (w8a16): 6.56 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| Mobile-VIT | ONNX | float | Snapdragon® X Elite | 4.702 ms | 12 - 12 MB | NPU
| Mobile-VIT | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 3.142 ms | 0 - 168 MB | NPU
| Mobile-VIT | ONNX | float | Qualcomm® QCS8550 (Proxy) | 4.544 ms | 0 - 133 MB | NPU
| Mobile-VIT | ONNX | float | Qualcomm® QCS9075 | 5.619 ms | 1 - 4 MB | NPU
| Mobile-VIT | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 2.534 ms | 0 - 139 MB | NPU
| Mobile-VIT | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 2.145 ms | 0 - 139 MB | NPU
| Mobile-VIT | ONNX | w8a16 | Snapdragon® X Elite | 16.897 ms | 16 - 16 MB | NPU
| Mobile-VIT | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 15.32 ms | 13 - 189 MB | NPU
| Mobile-VIT | ONNX | w8a16 | Qualcomm® QCS6490 | 339.344 ms | 69 - 73 MB | CPU
| Mobile-VIT | ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 18.861 ms | 7 - 17 MB | NPU
| Mobile-VIT | ONNX | w8a16 | Qualcomm® QCS9075 | 20.873 ms | 12 - 14 MB | NPU
| Mobile-VIT | ONNX | w8a16 | Qualcomm® QCM6690 | 149.372 ms | 63 - 74 MB | CPU
| Mobile-VIT | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 12.17 ms | 13 - 152 MB | NPU
| Mobile-VIT | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 131.756 ms | 63 - 73 MB | CPU
| Mobile-VIT | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 11.754 ms | 12 - 151 MB | NPU
| Mobile-VIT | QNN_DLC | float | Snapdragon® X Elite | 3.849 ms | 1 - 1 MB | NPU
| Mobile-VIT | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 2.456 ms | 0 - 98 MB | NPU
| Mobile-VIT | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 9.874 ms | 1 - 59 MB | NPU
| Mobile-VIT | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 3.48 ms | 0 - 13 MB | NPU
| Mobile-VIT | QNN_DLC | float | Qualcomm® SA8775P | 4.259 ms | 1 - 61 MB | NPU
| Mobile-VIT | QNN_DLC | float | Qualcomm® QCS9075 | 4.47 ms | 1 - 3 MB | NPU
| Mobile-VIT | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 5.699 ms | 0 - 94 MB | NPU
| Mobile-VIT | QNN_DLC | float | Qualcomm® SA7255P | 9.874 ms | 1 - 59 MB | NPU
| Mobile-VIT | QNN_DLC | float | Qualcomm® SA8295P | 6.44 ms | 1 - 63 MB | NPU
| Mobile-VIT | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.907 ms | 1 - 68 MB | NPU
| Mobile-VIT | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.611 ms | 1 - 73 MB | NPU
| Mobile-VIT | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 2.58 ms | 0 - 108 MB | NPU
| Mobile-VIT | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 10.21 ms | 0 - 85 MB | NPU
| Mobile-VIT | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 3.652 ms | 0 - 19 MB | NPU
| Mobile-VIT | TFLITE | float | Qualcomm® SA8775P | 4.461 ms | 0 - 80 MB | NPU
| Mobile-VIT | TFLITE | float | Qualcomm® QCS9075 | 4.555 ms | 0 - 15 MB | NPU
| Mobile-VIT | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 6.017 ms | 0 - 94 MB | NPU
| Mobile-VIT | TFLITE | float | Qualcomm® SA7255P | 10.21 ms | 0 - 85 MB | NPU
| Mobile-VIT | TFLITE | float | Qualcomm® SA8295P | 6.691 ms | 0 - 71 MB | NPU
| Mobile-VIT | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 2.016 ms | 0 - 74 MB | NPU
| Mobile-VIT | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.644 ms | 0 - 84 MB | NPU

## License
* The license for the original implementation of Mobile-VIT can be found
  [here](https://github.com/pytorch/vision/blob/main/LICENSE).

## References
* [MOBILEVIT: LIGHT-WEIGHT, GENERAL-PURPOSE, AND MOBILE-FRIENDLY VISION TRANSFORMER](https://arxiv.org/abs/2110.02178)
* [Source Model Implementation](https://github.com/apple/ml-cvnets)

## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).