---
library_name: pytorch
license: other
tags:
- real_time
- android
pipeline_tag: object-detection

---

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

# Yolo-X: Optimized for Qualcomm Devices

YoloX is a machine learning model that predicts bounding boxes and classes of objects in an image.

This is based on the implementation of Yolo-X found [here](https://github.com/Megvii-BaseDetection/YOLOX/).
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/main/src/qai_hub_models/models/yolox) 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 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-onnx-float.zip)
| ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-onnx-w8a16.zip)
| ONNX | w8a8 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-onnx-w8a8.zip)
| ONNX | w8a8_mixed_int16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-onnx-w8a8_mixed_int16.zip)
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-qnn_dlc-w8a16.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-qnn_dlc-w8a8.zip)
| QNN_DLC | w8a8_mixed_int16 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-qnn_dlc-w8a8_mixed_int16.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.51.0/yolox-tflite-w8a8.zip)

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


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/yolox) 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 [Yolo-X on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/yolox) for usage instructions.

## Model Details

**Model Type:** Model_use_case.object_detection

**Model Stats:**
- Model checkpoint: YoloX Small
- Input resolution: 640x640
- Number of parameters: 8.98M
- Model size (float): 34.3 MB
- Model size (w8a16): 9.53 MB
- Model size (w8a8): 8.96 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| Yolo-X | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.134 ms | 5 - 166 MB | NPU
| Yolo-X | ONNX | float | Snapdragon® X2 Elite | 7.616 ms | 14 - 14 MB | NPU
| Yolo-X | ONNX | float | Snapdragon® X Elite | 13.412 ms | 14 - 14 MB | NPU
| Yolo-X | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 8.977 ms | 5 - 192 MB | NPU
| Yolo-X | ONNX | float | Qualcomm® QCS8550 (Proxy) | 13.662 ms | 0 - 17 MB | NPU
| Yolo-X | ONNX | float | Qualcomm® QCS9075 | 16.77 ms | 5 - 8 MB | NPU
| Yolo-X | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 7.684 ms | 1 - 162 MB | NPU
| Yolo-X | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 10.864 ms | 0 - 203 MB | NPU
| Yolo-X | ONNX | w8a16 | Snapdragon® X2 Elite | 11.124 ms | 8 - 8 MB | NPU
| Yolo-X | ONNX | w8a16 | Snapdragon® X Elite | 14.605 ms | 7 - 7 MB | NPU
| Yolo-X | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 9.933 ms | 3 - 242 MB | NPU
| Yolo-X | ONNX | w8a16 | Qualcomm® QCS6490 | 776.777 ms | 113 - 116 MB | CPU
| Yolo-X | ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 13.926 ms | 0 - 11 MB | NPU
| Yolo-X | ONNX | w8a16 | Qualcomm® QCS9075 | 14.751 ms | 2 - 5 MB | NPU
| Yolo-X | ONNX | w8a16 | Qualcomm® QCM6690 | 421.182 ms | 138 - 148 MB | CPU
| Yolo-X | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 8.659 ms | 1 - 198 MB | NPU
| Yolo-X | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 394.704 ms | 112 - 122 MB | CPU
| Yolo-X | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 8.097 ms | 1 - 59 MB | NPU
| Yolo-X | ONNX | w8a8 | Snapdragon® X2 Elite | 8.123 ms | 8 - 8 MB | NPU
| Yolo-X | ONNX | w8a8 | Snapdragon® X Elite | 9.205 ms | 8 - 8 MB | NPU
| Yolo-X | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 6.237 ms | 0 - 89 MB | NPU
| Yolo-X | ONNX | w8a8 | Qualcomm® QCS6490 | 109.431 ms | 48 - 57 MB | CPU
| Yolo-X | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 8.336 ms | 1 - 17 MB | NPU
| Yolo-X | ONNX | w8a8 | Qualcomm® QCS9075 | 9.372 ms | 1 - 4 MB | NPU
| Yolo-X | ONNX | w8a8 | Qualcomm® QCM6690 | 98.645 ms | 38 - 48 MB | CPU
| Yolo-X | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 5.79 ms | 1 - 53 MB | NPU
| Yolo-X | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 89.993 ms | 48 - 58 MB | CPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Snapdragon® 8 Elite Gen 5 Mobile | 26.723 ms | 10 - 737 MB | NPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Snapdragon® X2 Elite | 28.421 ms | 19 - 19 MB | NPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Snapdragon® X Elite | 36.423 ms | 16 - 16 MB | NPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Snapdragon® 8 Gen 3 Mobile | 27.607 ms | 12 - 835 MB | NPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Qualcomm® QCS6490 | 317.339 ms | 104 - 113 MB | CPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Qualcomm® QCS8550 (Proxy) | 39.536 ms | 9 - 14 MB | NPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Qualcomm® QCS9075 | 38.535 ms | 12 - 15 MB | NPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Qualcomm® QCM6690 | 207.438 ms | 94 - 105 MB | CPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Snapdragon® 8 Elite For Galaxy Mobile | 24.944 ms | 10 - 727 MB | NPU
| Yolo-X | ONNX | w8a8_mixed_int16 | Snapdragon® 7 Gen 4 Mobile | 191.588 ms | 111 - 121 MB | CPU
| Yolo-X | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 1.926 ms | 2 - 84 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 2.636 ms | 2 - 2 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Snapdragon® X Elite | 6.084 ms | 2 - 2 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 3.63 ms | 2 - 107 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® QCS6490 | 16.881 ms | 2 - 6 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® QCS8275 (Proxy) | 11.663 ms | 0 - 71 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® QCS8550 (Proxy) | 5.519 ms | 2 - 4 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® SA8775P | 6.141 ms | 1 - 74 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® QCS9075 | 5.994 ms | 0 - 4 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® QCM6690 | 50.908 ms | 2 - 196 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® QCS8450 (Proxy) | 6.917 ms | 2 - 107 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® SA7255P | 11.663 ms | 0 - 71 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Qualcomm® SA8295P | 7.625 ms | 0 - 70 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 2.612 ms | 2 - 79 MB | NPU
| Yolo-X | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 7.006 ms | 2 - 191 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.905 ms | 1 - 52 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 1.206 ms | 1 - 1 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Snapdragon® X Elite | 2.735 ms | 1 - 1 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.645 ms | 3 - 73 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 7.767 ms | 3 - 6 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® QCS8275 (Proxy) | 5.83 ms | 1 - 49 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 2.464 ms | 1 - 29 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® SA8775P | 2.814 ms | 1 - 52 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 3.326 ms | 1 - 4 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 22.405 ms | 2 - 163 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® QCS8450 (Proxy) | 2.977 ms | 1 - 76 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® SA7255P | 5.83 ms | 1 - 49 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Qualcomm® SA8295P | 3.79 ms | 1 - 48 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.176 ms | 1 - 52 MB | NPU
| Yolo-X | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 2.784 ms | 1 - 50 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Snapdragon® 8 Elite Gen 5 Mobile | 1.374 ms | 1 - 70 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Snapdragon® X2 Elite | 1.855 ms | 1 - 1 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Snapdragon® X Elite | 4.089 ms | 1 - 1 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Snapdragon® 8 Gen 3 Mobile | 2.423 ms | 1 - 91 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Qualcomm® QCS8275 (Proxy) | 8.128 ms | 1 - 59 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Qualcomm® QCS8550 (Proxy) | 3.695 ms | 1 - 3 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Qualcomm® SA8775P | 4.119 ms | 1 - 63 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Qualcomm® QCS9075 | 4.529 ms | 1 - 4 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Qualcomm® QCM6690 | 34.551 ms | 1 - 180 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Qualcomm® SA7255P | 8.128 ms | 1 - 59 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Snapdragon® 8 Elite For Galaxy Mobile | 1.771 ms | 1 - 67 MB | NPU
| Yolo-X | QNN_DLC | w8a8_mixed_int16 | Snapdragon® 7 Gen 4 Mobile | 4.62 ms | 1 - 177 MB | NPU
| Yolo-X | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.912 ms | 0 - 50 MB | NPU
| Yolo-X | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.639 ms | 0 - 75 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® QCS6490 | 7.475 ms | 0 - 12 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® QCS8275 (Proxy) | 5.873 ms | 0 - 45 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 2.438 ms | 0 - 2 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® SA8775P | 2.912 ms | 0 - 51 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® QCS9075 | 3.37 ms | 0 - 12 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® QCM6690 | 21.199 ms | 0 - 162 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 2.962 ms | 0 - 71 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® SA7255P | 5.873 ms | 0 - 45 MB | NPU
| Yolo-X | TFLITE | w8a8 | Qualcomm® SA8295P | 3.84 ms | 0 - 44 MB | NPU
| Yolo-X | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.202 ms | 0 - 45 MB | NPU
| Yolo-X | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 2.766 ms | 0 - 46 MB | NPU

## License
* The license for the original implementation of Yolo-X can be found
  [here](https://github.com/Megvii-BaseDetection/YOLOX/blob/main/LICENSE).

## References
* [YOLOX: Exceeding YOLO Series in 2021](https://github.com/Megvii-BaseDetection/YOLOX/blob/main/README.md)
* [Source Model Implementation](https://github.com/Megvii-BaseDetection/YOLOX/)

## 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).