Hugging Face's logo

qualcomm
/
ResNeXt50

Image Classification
PyTorch
backbone
bu_auto
android
Model card Files
xet
 Community
ResNeXt50
File size: 12,500 Bytes
1ca3652
 
dc9ebd3
1ca3652
 
5c3ad4e
1ca3652
a7ea068
1ca3652
 
 
e70eebf
1ca3652
f5579c2
e882320
1ca3652
 
f5579c2
e0a33d7
f5579c2
 
 
 
 
 
 
 
 
 
 
 
e0a33d7
 
 
 
 
 
f5579c2
 
 
 
 
 
e0a33d7
f5579c2
 
 
 
 
 
e0a33d7
f5579c2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e0a33d7
 
 
 
 
 
 
 
 
 
9ef3c02
e0a33d7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9ef3c02
e0a33d7
 
 
 
 
 
aa829d1
1ca3652
98c2ea2
 
aa829d1
1ca3652
 
 
 
 
66f1506
1ca3652
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
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
---
library_name: pytorch
license: other
tags:
- backbone
- bu_auto
- android
pipeline_tag: image-classification

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext50/web-assets/model_demo.png)

# ResNeXt50: Optimized for Qualcomm Devices

ResNeXt50 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 ResNeXt50 found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/resnet.py).
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/v0.59.0/src/qai_hub_models/models/resnext50) 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 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext50/releases/v0.59.0/resnext50-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext50/releases/v0.59.0/resnext50-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext50/releases/v0.59.0/resnext50-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext50/releases/v0.59.0/resnext50-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext50/releases/v0.59.0/resnext50-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext50/releases/v0.59.0/resnext50-tflite-w8a8.zip)

For more device-specific assets and performance metrics, visit **[ResNeXt50 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/resnext50)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/resnext50) 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 [ResNeXt50 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/resnext50) for usage instructions.

## Model Details

**Model Type:** Model_use_case.image_classification

**Model Stats:**
- Model checkpoint: Imagenet
- Input resolution: 224x224
- Number of parameters: 25.0M
- Model size (float): 95.4 MB
- Model size (w8a8): 24.8 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| ResNeXt50 | ONNX | float | Snapdragon® X2 Elite | 1.114 ms | 2 - 2 MB | NPU
| ResNeXt50 | ONNX | float | Snapdragon® X Elite | 2.234 ms | 50 - 50 MB | NPU
| ResNeXt50 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 1.63 ms | 0 - 136 MB | NPU
| ResNeXt50 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 3.484 ms | 0 - 115 MB | NPU
| ResNeXt50 | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.268 ms | 0 - 57 MB | NPU
| ResNeXt50 | ONNX | float | Qualcomm® QCS8450 | 3.484 ms | 0 - 115 MB | NPU
| ResNeXt50 | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 3.444 ms | 0 - 4 MB | NPU
| ResNeXt50 | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 2.234 ms | 50 - 50 MB | NPU
| ResNeXt50 | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 1.384 ms | 0 - 73 MB | NPU
| ResNeXt50 | ONNX | float | Snapdragon® 8 Elite Mobile | 1.384 ms | 0 - 73 MB | NPU
| ResNeXt50 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.145 ms | 0 - 74 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Snapdragon® X2 Elite | 0.51 ms | 1 - 1 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Snapdragon® X Elite | 1.091 ms | 26 - 26 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.816 ms | 0 - 108 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.481 ms | 0 - 107 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 4.072 ms | 0 - 3 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.1 ms | 0 - 185 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Qualcomm® QCS8450 | 1.481 ms | 0 - 107 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.323 ms | 0 - 3 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 1.091 ms | 26 - 26 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.702 ms | 0 - 76 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 0.702 ms | 0 - 76 MB | NPU
| ResNeXt50 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.617 ms | 0 - 80 MB | NPU
| ResNeXt50 | QNN_DLC | float | Snapdragon® X2 Elite | 1.44 ms | 1 - 1 MB | NPU
| ResNeXt50 | QNN_DLC | float | Snapdragon® X Elite | 2.66 ms | 1 - 1 MB | NPU
| ResNeXt50 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 1.789 ms | 0 - 133 MB | NPU
| ResNeXt50 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 4.018 ms | 0 - 114 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8275 | 12.017 ms | 1 - 72 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.523 ms | 1 - 3 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® SA8775P | 3.818 ms | 1 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® SA8650P | 3.818 ms | 1 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® SA8255P | 3.818 ms | 1 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® QCS8450 | 4.018 ms | 0 - 114 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 3.633 ms | 1 - 3 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 2.66 ms | 1 - 1 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 1.485 ms | 1 - 73 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® SA7255P | 12.017 ms | 1 - 72 MB | NPU
| ResNeXt50 | QNN_DLC | float | Qualcomm® SA8295P | 4.098 ms | 0 - 51 MB | NPU
| ResNeXt50 | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 1.485 ms | 1 - 73 MB | NPU
| ResNeXt50 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.179 ms | 1 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.612 ms | 0 - 0 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Snapdragon® X Elite | 1.224 ms | 0 - 0 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.781 ms | 0 - 101 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.46 ms | 0 - 101 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 4.188 ms | 0 - 2 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8275 | 2.459 ms | 0 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.077 ms | 0 - 2 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® SA8775P | 1.456 ms | 0 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® SA8650P | 1.456 ms | 0 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® SA8255P | 1.456 ms | 0 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 1.46 ms | 0 - 101 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.296 ms | 2 - 4 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 1.224 ms | 0 - 0 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 7.757 ms | 0 - 196 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.376 ms | 0 - 77 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.637 ms | 0 - 76 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® SA7255P | 2.459 ms | 0 - 74 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 1.772 ms | 0 - 71 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 0.637 ms | 0 - 76 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.543 ms | 0 - 76 MB | NPU
| ResNeXt50 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.376 ms | 0 - 77 MB | NPU
| ResNeXt50 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 1.76 ms | 0 - 171 MB | NPU
| ResNeXt50 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 4.0 ms | 0 - 151 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 12.028 ms | 0 - 112 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.472 ms | 0 - 2 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® SA8775P | 3.828 ms | 0 - 113 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® SA8650P | 3.828 ms | 0 - 113 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® SA8255P | 3.828 ms | 0 - 113 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® QCS8450 | 4.0 ms | 0 - 151 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 3.771 ms | 0 - 52 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 1.434 ms | 0 - 116 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® SA7255P | 12.028 ms | 0 - 112 MB | NPU
| ResNeXt50 | TFLITE | float | Qualcomm® SA8295P | 4.14 ms | 0 - 87 MB | NPU
| ResNeXt50 | TFLITE | float | Snapdragon® 8 Elite Mobile | 1.434 ms | 0 - 116 MB | NPU
| ResNeXt50 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.194 ms | 0 - 118 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.668 ms | 0 - 102 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.285 ms | 0 - 99 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 3.834 ms | 0 - 27 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8275 | 2.209 ms | 0 - 70 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.896 ms | 0 - 2 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® SA8775P | 1.268 ms | 0 - 72 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® SA8650P | 1.268 ms | 0 - 72 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® SA8255P | 1.268 ms | 0 - 72 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® QCS8450 | 1.285 ms | 0 - 99 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.087 ms | 0 - 27 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 7.107 ms | 0 - 198 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.219 ms | 0 - 70 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.579 ms | 0 - 73 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® SA7255P | 2.209 ms | 0 - 70 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Qualcomm® SA8295P | 1.568 ms | 0 - 65 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 0.579 ms | 0 - 73 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.49 ms | 0 - 68 MB | NPU
| ResNeXt50 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.219 ms | 0 - 70 MB | NPU

## License
* The license for the original implementation of ResNeXt50 can be found
  [here](https://github.com/pytorch/vision/blob/main/LICENSE).

## References
* [Aggregated Residual Transformations for Deep Neural Networks](https://arxiv.org/abs/1611.05431)
* [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/resnet.py)

## 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).