README.md

---
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/main/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 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.50.2/deeplab_xception-onnx-float.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/deeplab_xception/releases/v0.50.2/deeplab_xception-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.50.2/deeplab_xception-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.50.2/deeplab_xception-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.43, TFLite 2.19.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.50.2/deeplab_xception-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.43, TFLite 2.19.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplab_xception/releases/v0.50.2/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/main/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/main/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® 8 Elite Gen 5 Mobile | 8.881 ms | 3 - 255 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® X2 Elite | 10.453 ms | 86 - 86 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® X Elite | 22.956 ms | 85 - 85 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 17.0 ms | 0 - 367 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS8550 (Proxy) | 22.43 ms | 0 - 96 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS9075 | 33.553 ms | 3 - 8 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 13.26 ms | 1 - 242 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 3.058 ms | 0 - 293 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® X2 Elite | 3.348 ms | 44 - 44 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® X Elite | 7.998 ms | 44 - 44 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 5.275 ms | 0 - 389 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS6490 | 318.533 ms | 92 - 119 MB | CPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 7.438 ms | 0 - 51 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS9075 | 7.628 ms | 0 - 4 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCM6690 | 244.569 ms | 59 - 73 MB | CPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 4.211 ms | 0 - 283 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 229.761 ms | 91 - 105 MB | CPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.525 ms | 3 - 258 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® X2 Elite | 11.481 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® X Elite | 22.739 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 16.008 ms | 2 - 367 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 21.914 ms | 3 - 5 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA8775P | 30.623 ms | 0 - 244 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS9075 | 36.016 ms | 5 - 10 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 46.323 ms | 3 - 363 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA8295P | 39.001 ms | 0 - 241 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 12.179 ms | 3 - 254 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 3.26 ms | 1 - 271 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 3.869 ms | 1 - 1 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® X Elite | 8.53 ms | 1 - 1 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 5.55 ms | 1 - 373 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 33.811 ms | 1 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8275 (Proxy) | 20.336 ms | 1 - 282 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 7.888 ms | 1 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA8775P | 8.085 ms | 1 - 285 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 8.711 ms | 1 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 93.454 ms | 1 - 303 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8450 (Proxy) | 13.82 ms | 1 - 367 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA7255P | 20.336 ms | 1 - 282 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA8295P | 12.624 ms | 1 - 277 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 4.377 ms | 1 - 287 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 12.101 ms | 1 - 277 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.204 ms | 0 - 369 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 16.148 ms | 0 - 482 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 113.261 ms | 0 - 358 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 21.316 ms | 0 - 8 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8775P | 30.584 ms | 0 - 358 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS9075 | 35.556 ms | 0 - 93 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 45.665 ms | 0 - 473 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA7255P | 113.261 ms | 0 - 358 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8295P | 38.684 ms | 0 - 349 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 12.244 ms | 0 - 364 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 2.897 ms | 0 - 267 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 4.952 ms | 0 - 373 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS6490 | 31.564 ms | 0 - 58 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8275 (Proxy) | 18.929 ms | 0 - 285 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 6.935 ms | 0 - 3 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8775P | 7.234 ms | 0 - 284 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS9075 | 7.499 ms | 0 - 47 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCM6690 | 90.917 ms | 0 - 295 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 12.237 ms | 0 - 361 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA7255P | 18.929 ms | 0 - 285 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8295P | 11.612 ms | 0 - 273 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 3.905 ms | 0 - 273 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 11.136 ms | 0 - 268 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).