Hugging Face's logo

qualcomm
/
DeepLabXception

Image Segmentation
PyTorch
bu_auto
android
Model card Files
xet
 Community
DeepLabXception
File size: 13,514 Bytes
0d3a78b
 
 
 
aba40f5
0d3a78b
 
 
 
 
 
 
69e2e85
0d3a78b
 
 
69e2e85
052606d
69e2e85
 
 
 
 
 
 
 
 
 
 
 
052606d
 
 
 
 
 
69e2e85
 
 
 
 
 
052606d
69e2e85
 
 
 
 
 
052606d
69e2e85
 
 
 
 
 
 
 
 
 
 
 
 
 
 
052606d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0d3a78b
 
 
 
 
 
 
 
 
 
 
 
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
173
174
---
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/v0.58.0/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/v0.58.0/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/v0.58.0/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/v0.58.0/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/v0.58.0/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/v0.58.0/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/v0.58.0/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/v0.58.0/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/v0.58.0/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 | 10.45 ms | 3 - 3 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® X Elite | 22.92 ms | 86 - 86 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 17.187 ms | 3 - 361 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 50.62 ms | 4 - 357 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 22.867 ms | 0 - 95 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® QCS8450 | 50.62 ms | 4 - 357 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 31.693 ms | 3 - 8 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.983 ms | 2 - 254 MB | NPU
| DeepLabXception | ONNX | float | Snapdragon® 8 Elite Mobile | 13.232 ms | 2 - 250 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 13.232 ms | 2 - 250 MB | NPU
| DeepLabXception | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 22.92 ms | 86 - 86 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® X2 Elite | 3.409 ms | 2 - 2 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® X Elite | 7.772 ms | 44 - 44 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 5.28 ms | 0 - 385 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 12.49 ms | 1 - 379 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 45.724 ms | 0 - 4 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 7.455 ms | 1 - 4 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® QCS8450 | 12.49 ms | 1 - 379 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 4.232 ms | 0 - 289 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 11.747 ms | 1 - 280 MB | NPU
| DeepLabXception | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 3.163 ms | 0 - 298 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 7.334 ms | 0 - 4 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 91.646 ms | 1 - 309 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 11.747 ms | 1 - 280 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 4.232 ms | 0 - 289 MB | NPU
| DeepLabXception | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 7.772 ms | 44 - 44 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® X2 Elite | 11.514 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® X Elite | 22.948 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 16.134 ms | 0 - 352 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 45.828 ms | 3 - 355 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8275 | 112.981 ms | 1 - 244 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 22.061 ms | 3 - 5 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA8775P | 30.632 ms | 0 - 244 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA8650P | 30.632 ms | 0 - 244 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA8255P | 30.632 ms | 0 - 244 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® QCS8450 | 45.828 ms | 3 - 355 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 31.45 ms | 3 - 8 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.645 ms | 3 - 260 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA7255P | 112.981 ms | 1 - 244 MB | NPU
| DeepLabXception | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 12.183 ms | 0 - 250 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® SA8295P | 39.12 ms | 0 - 240 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 12.183 ms | 0 - 250 MB | NPU
| DeepLabXception | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 22.948 ms | 3 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 3.858 ms | 1 - 1 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® X Elite | 8.573 ms | 1 - 1 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 5.556 ms | 1 - 370 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 13.676 ms | 1 - 365 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 46.651 ms | 3 - 5 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8275 | 20.339 ms | 1 - 283 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 7.874 ms | 1 - 3 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA8775P | 8.076 ms | 1 - 286 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA8650P | 8.076 ms | 1 - 286 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA8255P | 8.076 ms | 1 - 286 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 13.676 ms | 1 - 365 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 4.367 ms | 1 - 289 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 12.807 ms | 1 - 277 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 3.281 ms | 1 - 282 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 7.775 ms | 3 - 5 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 93.567 ms | 1 - 302 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA7255P | 20.339 ms | 1 - 283 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® SA8295P | 12.726 ms | 0 - 279 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 12.807 ms | 1 - 277 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 4.367 ms | 1 - 289 MB | NPU
| DeepLabXception | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 8.573 ms | 1 - 1 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 16.201 ms | 0 - 455 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 45.404 ms | 1 - 451 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8275 | 113.151 ms | 1 - 343 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 21.309 ms | 0 - 3 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8775P | 30.633 ms | 0 - 344 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8650P | 30.633 ms | 0 - 344 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8255P | 30.633 ms | 0 - 344 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® QCS8450 | 45.404 ms | 1 - 451 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 31.078 ms | 0 - 93 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.292 ms | 0 - 359 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA7255P | 113.151 ms | 1 - 343 MB | NPU
| DeepLabXception | TFLITE | float | Snapdragon® 8 Elite Mobile | 12.237 ms | 0 - 355 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® SA8295P | 38.739 ms | 0 - 336 MB | NPU
| DeepLabXception | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 12.237 ms | 0 - 355 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 4.989 ms | 0 - 369 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 12.409 ms | 0 - 359 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 44.859 ms | 0 - 58 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8275 | 19.034 ms | 0 - 284 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.952 ms | 0 - 2 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8775P | 7.275 ms | 0 - 285 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8650P | 7.275 ms | 0 - 285 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8255P | 7.275 ms | 0 - 285 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® QCS8450 | 12.409 ms | 0 - 359 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 3.908 ms | 1 - 269 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 11.178 ms | 0 - 265 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 2.914 ms | 0 - 285 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 6.963 ms | 0 - 47 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 91.018 ms | 0 - 291 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA7255P | 19.034 ms | 0 - 284 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® SA8295P | 11.616 ms | 0 - 276 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 11.178 ms | 0 - 265 MB | NPU
| DeepLabXception | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 3.908 ms | 1 - 269 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).