---
library_name: pytorch
license: other
tags:
- bu_auto
- android
pipeline_tag: image-segmentation

---

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

# DeepLabXception: Optimized for Qualcomm Devices

DeepLabXception is a semantic segmentation model supporting multiple backbones like ResNet-101 and Xception, with flexible dataset compatibility including COCO, VOC, and Cityscapes.

This is based on the implementation of DeepLabXception found [here](https://github.com/LikeLy-Journey/SegmenTron).
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/v0.57.0/src/qai_hub_models/models/deeplab_xception) 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 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.57.0/deeplab_xception-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.57.0/deeplab_xception-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/deeplab_xception/releases/v0.57.0/deeplab_xception-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/deeplab_xception/releases/v0.57.0/deeplab_xception-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/deeplab_xception/releases/v0.57.0/deeplab_xception-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.57.0/deeplab_xception-tflite-w8a8.zip)

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


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.57.0/src/qai_hub_models/models/deeplab_xception) 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 [DeepLabXception on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.57.0/src/qai_hub_models/models/deeplab_xception) for usage instructions.

## Model Details

**Model Type:** Model_use_case.semantic_segmentation

**Model Stats:**
- Model checkpoint: COCO_WITH_VOC_LABELS_V1
- Input resolution: 480x520
- Number of output classes: 21
- Number of parameters: 41.26M
- Model size (float): 158 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| DeepLabXception | ONNX | float | Snapdragon® X2 Elite | 10.479 ms | 210 - 210 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® X Elite | 23.154 ms | 146 - 146 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 17.211 ms | 14 - 360 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 50.153 ms | 4 - 345 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS8550 (Proxy) | 22.791 ms | 0 - 97 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS8450 | 50.153 ms | 4 - 345 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Elite Mobile | 13.28 ms | 1 - 250 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.897 ms | 2 - 251 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS9075 | 33.357 ms | 3 - 48 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS8750 | 13.28 ms | 1 - 250 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS7181 | 23.154 ms | 146 - 146 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® X2 Elite | 3.4 ms | 180 - 180 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® X Elite | 7.76 ms | 181 - 181 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 5.264 ms | 0 - 373 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 12.409 ms | 1 - 369 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS6490 | 32.732 ms | 1 - 46 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 7.46 ms | 0 - 52 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS8450 | 12.409 ms | 1 - 369 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 3.152 ms | 0 - 293 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 11.762 ms | 1 - 278 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCM6690 | 91.442 ms | 0 - 300 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS9075 | 7.573 ms | 1 - 46 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 4.226 ms | 0 - 289 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS7790 | 11.762 ms | 1 - 278 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS8750 | 4.226 ms | 0 - 289 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS7181 | 7.76 ms | 181 - 181 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® X2 Elite | 11.549 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® X Elite | 22.973 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 16.144 ms | 2 - 356 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 45.687 ms | 3 - 356 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8275 | 112.974 ms | 2 - 246 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 21.918 ms | 3 - 5 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8450 | 45.687 ms | 3 - 356 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 12.193 ms | 0 - 249 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA8295P | 39.103 ms | 1 - 241 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.551 ms | 2 - 260 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA7255P | 112.974 ms | 2 - 246 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS9075 | 35.844 ms | 3 - 8 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8750 | 12.193 ms | 0 - 249 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS7181 | 22.973 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 3.883 ms | 1 - 1 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® X Elite | 8.575 ms | 1 - 1 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 5.553 ms | 1 - 371 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 13.734 ms | 1 - 369 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 33.678 ms | 3 - 5 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8275 | 20.381 ms | 1 - 283 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 7.909 ms | 1 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 13.734 ms | 1 - 369 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 3.263 ms | 1 - 281 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 12.469 ms | 1 - 277 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 93.37 ms | 1 - 303 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 8.623 ms | 1 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA7255P | 20.381 ms | 1 - 283 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 4.38 ms | 1 - 290 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA8295P | 12.697 ms | 1 - 280 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS7790 | 12.469 ms | 1 - 277 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8750 | 4.38 ms | 1 - 290 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS7181 | 8.575 ms | 1 - 1 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 16.16 ms | 0 - 455 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 45.73 ms | 0 - 451 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8275 | 113.136 ms | 1 - 344 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 21.578 ms | 0 - 2 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8775P | 206.286 ms | 21 - 53 MB | GPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8650P | 206.286 ms | 21 - 53 MB | GPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8255P | 206.286 ms | 21 - 53 MB | GPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8450 | 45.73 ms | 0 - 451 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Elite Mobile | 12.193 ms | 0 - 356 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8295P | 38.738 ms | 0 - 337 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.196 ms | 0 - 359 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA7255P | 113.136 ms | 1 - 344 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS9075 | 35.078 ms | 0 - 93 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8750 | 12.193 ms | 0 - 356 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 4.979 ms | 0 - 366 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 12.122 ms | 0 - 358 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS6490 | 31.782 ms | 0 - 58 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8275 | 19.031 ms | 0 - 284 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 6.98 ms | 0 - 2 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8775P | 208.882 ms | 28 - 61 MB | GPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8650P | 208.882 ms | 28 - 61 MB | GPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8255P | 208.882 ms | 28 - 61 MB | GPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8450 | 12.122 ms | 0 - 358 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 2.896 ms | 0 - 286 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 11.229 ms | 0 - 266 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCM6690 | 90.99 ms | 0 - 291 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS9075 | 7.473 ms | 0 - 47 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA7255P | 19.031 ms | 0 - 284 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 3.909 ms | 0 - 276 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8295P | 11.635 ms | 0 - 276 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS7790 | 11.229 ms | 0 - 266 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8750 | 3.909 ms | 0 - 276 MB | NPU

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

## References
* [Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation](https://arxiv.org/abs/1802.02611)
* [Source Model Implementation](https://github.com/LikeLy-Journey/SegmenTron)

## 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).