---
library_name: pytorch
license: bsd-3-clause
tags:
- backbone
- bu_auto
- android
pipeline_tag: image-classification

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext101/web-assets/model_demo.png)

# ResNeXt101: Optimized for Qualcomm Devices

ResNeXt101 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 ResNeXt101 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/v0.63.0/src/qai_hub_models/models/resnext101) 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 2.50, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext101/releases/v0.63.0/resnext101-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.50, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext101/releases/v0.63.0/resnext101-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext101/releases/v0.63.0/resnext101-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext101/releases/v0.63.0/resnext101-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext101/releases/v0.63.0/resnext101-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/resnext101/releases/v0.63.0/resnext101-tflite-w8a8.zip)

For more device-specific assets and performance metrics, visit **[ResNeXt101 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/resnext101)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.63.0/src/qai_hub_models/models/resnext101) 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 [ResNeXt101 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.63.0/src/qai_hub_models/models/resnext101) for usage instructions.

## Model Details

**Model Type:** Model_use_case.image_classification

**Model Stats:**
- Input resolution: 224x224
- Model checkpoint: Imagenet
- Model size (float): 338 MB
- Model size (w8a8): 87.3 MB
- Number of parameters: 88.7M

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| ResNeXt101 | ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 3.092 ms | 1 - 190 MB | NPU
| ResNeXt101 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 3.921 ms | 0 - 184 MB | NPU
| ResNeXt101 | ONNX | float | Snapdragon® X2 Elite | 3.083 ms | 2 - 2 MB | NPU
| ResNeXt101 | ONNX | float | Snapdragon® X Elite | 6.253 ms | 173 - 173 MB | NPU
| ResNeXt101 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 4.486 ms | 0 - 366 MB | NPU
| ResNeXt101 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 9.266 ms | 0 - 315 MB | NPU
| ResNeXt101 | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 10.864 ms | 0 - 5 MB | NPU
| ResNeXt101 | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.18 ms | 1 - 3 MB | NPU
| ResNeXt101 | ONNX | float | Qualcomm® QCS8450 | 9.266 ms | 0 - 315 MB | NPU
| ResNeXt101 | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 9.696 ms | 0 - 4 MB | NPU
| ResNeXt101 | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 6.253 ms | 173 - 173 MB | NPU
| ResNeXt101 | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 3.921 ms | 0 - 184 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 1.595 ms | 0 - 227 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 2.086 ms | 0 - 223 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Snapdragon® X2 Elite | 1.439 ms | 1 - 1 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Snapdragon® X Elite | 2.846 ms | 88 - 88 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 2.118 ms | 0 - 269 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 3.386 ms | 0 - 271 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 9.815 ms | 0 - 3 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 3.067 ms | 0 - 4 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.869 ms | 0 - 3 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® QCS8450 | 3.386 ms | 0 - 271 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 3.259 ms | 0 - 4 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 2.846 ms | 88 - 88 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 37.653 ms | 0 - 361 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 3.918 ms | 0 - 253 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 2.086 ms | 0 - 223 MB | NPU
| ResNeXt101 | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.918 ms | 0 - 253 MB | NPU
| ResNeXt101 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 3.122 ms | 1 - 184 MB | NPU
| ResNeXt101 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 3.98 ms | 0 - 183 MB | NPU
| ResNeXt101 | QNN_DLC | float | Snapdragon® X2 Elite | 3.483 ms | 1 - 1 MB | NPU
| ResNeXt101 | QNN_DLC | float | Snapdragon® X Elite | 6.746 ms | 1 - 1 MB | NPU
| ResNeXt101 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 4.61 ms | 0 - 364 MB | NPU
| ResNeXt101 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 9.811 ms | 0 - 311 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 10.823 ms | 1 - 4 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.411 ms | 1 - 3 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® SA8775P | 10.034 ms | 1 - 181 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® SA8650P | 10.034 ms | 1 - 181 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® SA8255P | 10.034 ms | 1 - 181 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® QCS8450 | 9.811 ms | 0 - 311 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 9.784 ms | 1 - 3 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 6.746 ms | 1 - 1 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 3.98 ms | 0 - 183 MB | NPU
| ResNeXt101 | QNN_DLC | float | Qualcomm® SA8295P | 10.806 ms | 1 - 137 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 1.516 ms | 0 - 225 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.792 ms | 0 - 220 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 1.678 ms | 0 - 0 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Snapdragon® X Elite | 3.05 ms | 0 - 0 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 2.089 ms | 0 - 265 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 3.357 ms | 0 - 268 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 10.046 ms | 0 - 2 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 3.024 ms | 0 - 3 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.837 ms | 0 - 271 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® SA8650P | 3.503 ms | 0 - 215 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® SA8255P | 3.503 ms | 0 - 215 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 3.357 ms | 0 - 268 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 3.219 ms | 2 - 4 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 3.05 ms | 0 - 0 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 33.95 ms | 0 - 361 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 3.946 ms | 0 - 254 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 1.792 ms | 0 - 220 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 4.04 ms | 0 - 225 MB | NPU
| ResNeXt101 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.946 ms | 0 - 254 MB | NPU
| ResNeXt101 | TFLITE | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 3.125 ms | 0 - 367 MB | NPU
| ResNeXt101 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 3.995 ms | 0 - 352 MB | NPU
| ResNeXt101 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 4.605 ms | 0 - 535 MB | NPU
| ResNeXt101 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 9.816 ms | 0 - 483 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 10.853 ms | 0 - 176 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.418 ms | 0 - 3 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® SA8775P | 10.144 ms | 0 - 357 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® SA8650P | 10.144 ms | 0 - 357 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® SA8255P | 10.144 ms | 0 - 357 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® QCS8450 | 9.816 ms | 0 - 483 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 9.859 ms | 0 - 175 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 3.995 ms | 0 - 352 MB | NPU
| ResNeXt101 | TFLITE | float | Qualcomm® SA7255P | 159.933 ms | 0 - 20 MB | GPU
| ResNeXt101 | TFLITE | float | Qualcomm® SA8295P | 10.787 ms | 0 - 309 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 1.462 ms | 0 - 219 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.718 ms | 0 - 211 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.973 ms | 0 - 256 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 3.211 ms | 0 - 253 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 9.289 ms | 0 - 89 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 2.858 ms | 0 - 90 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.628 ms | 0 - 2 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® SA8650P | 3.3 ms | 0 - 214 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® SA8255P | 3.3 ms | 0 - 214 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® QCS8450 | 3.211 ms | 0 - 253 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 3.012 ms | 0 - 89 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 26.014 ms | 0 - 354 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 3.863 ms | 0 - 246 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 1.718 ms | 0 - 211 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® SA7255P | 6.044 ms | 0 - 214 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Qualcomm® SA8295P | 3.892 ms | 0 - 219 MB | NPU
| ResNeXt101 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 3.863 ms | 0 - 246 MB | NPU

## License
* The license for the original implementation of ResNeXt101 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).