File size: 9,962 Bytes
8922adc c73c8e3 8922adc 2f02303 8922adc 5842d4a 8922adc 4625f72 8922adc 1b4c178 5842d4a 1b4c178 5842d4a 1b4c178 5842d4a 1b4c178 5842d4a 19c9bbb 5842d4a 19c9bbb 5842d4a 1b4c178 8922adc | 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 | ---
library_name: pytorch
license: mit
tags:
- bu_auto
- android
pipeline_tag: image-segmentation
---

# SalsaNext: Optimized for Qualcomm Devices
SalsaNext is a LiDAR-based model designed for efficient and accurate semantic segmentation.
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/v0.64.0/src/qai_hub_models/models/salsanext) 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 1.30.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/salsanext/releases/v0.64.0/salsanext-onnx-float.zip)
| ONNX | w8a16 | Universal | QAIRT 2.50, ONNX Runtime 1.30.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/salsanext/releases/v0.64.0/salsanext-onnx-w8a16.zip)
| QNN_DLC | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/salsanext/releases/v0.64.0/salsanext-qnn_dlc-float.zip)
| QNN_DLC | w8a16 | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/salsanext/releases/v0.64.0/salsanext-qnn_dlc-w8a16.zip)
| TFLITE | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/salsanext/releases/v0.64.0/salsanext-tflite-float.zip)
For more device-specific assets and performance metrics, visit **[SalsaNext on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/salsanext)**.
### Option 2: Export with Custom Configurations
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.64.0/src/qai_hub_models/models/salsanext) 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 [SalsaNext on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.64.0/src/qai_hub_models/models/salsanext) for usage instructions.
## Model Details
**Model Type:** Model_use_case.semantic_segmentation
**Model Stats:**
- Input resolution: 1x5x64x2048
- Model checkpoint: SalsaNext
- Model size (float): 25.7 MB
- Number of parameters: 6.71M
## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| SalsaNext | ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 15.488 ms | 23 - 266 MB | NPU
| SalsaNext | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 18.822 ms | 24 - 250 MB | NPU
| SalsaNext | ONNX | float | Snapdragon® X2 Elite | 14.454 ms | 35 - 35 MB | NPU
| SalsaNext | ONNX | float | Snapdragon® X Elite | 28.635 ms | 35 - 35 MB | NPU
| SalsaNext | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 21.315 ms | 25 - 327 MB | NPU
| SalsaNext | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 40.711 ms | 1 - 311 MB | NPU
| SalsaNext | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 38.856 ms | 22 - 27 MB | NPU
| SalsaNext | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 29.025 ms | 23 - 25 MB | NPU
| SalsaNext | ONNX | float | Qualcomm® QCS8450 | 40.711 ms | 1 - 311 MB | NPU
| SalsaNext | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 36.216 ms | 22 - 26 MB | NPU
| SalsaNext | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 28.635 ms | 35 - 35 MB | NPU
| SalsaNext | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 18.822 ms | 24 - 250 MB | NPU
| SalsaNext | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 21.962 ms | 12 - 357 MB | NPU
| SalsaNext | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 26.927 ms | 11 - 302 MB | NPU
| SalsaNext | ONNX | w8a16 | Snapdragon® X2 Elite | 17.913 ms | 18 - 18 MB | NPU
| SalsaNext | ONNX | w8a16 | Snapdragon® X Elite | 42.3 ms | 18 - 18 MB | NPU
| SalsaNext | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 33.846 ms | 7 - 372 MB | NPU
| SalsaNext | ONNX | w8a16 | Snapdragon® 8 Gen 1 Mobile | 62.112 ms | 12 - 431 MB | NPU
| SalsaNext | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 41.157 ms | 11 - 15 MB | NPU
| SalsaNext | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 43.457 ms | 10 - 22 MB | NPU
| SalsaNext | ONNX | w8a16 | Qualcomm® QCS8450 | 62.112 ms | 12 - 431 MB | NPU
| SalsaNext | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 44.055 ms | 12 - 15 MB | NPU
| SalsaNext | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 42.3 ms | 18 - 18 MB | NPU
| SalsaNext | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 26.927 ms | 11 - 302 MB | NPU
| SalsaNext | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 13.205 ms | 2 - 242 MB | NPU
| SalsaNext | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 18.144 ms | 3 - 227 MB | NPU
| SalsaNext | QNN_DLC | float | Snapdragon® X2 Elite | 14.509 ms | 3 - 3 MB | NPU
| SalsaNext | QNN_DLC | float | Snapdragon® X Elite | 27.966 ms | 3 - 3 MB | NPU
| SalsaNext | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 20.696 ms | 0 - 294 MB | NPU
| SalsaNext | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 40.494 ms | 0 - 304 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 38.212 ms | 3 - 18 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 28.06 ms | 3 - 6 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® SA8650P | 34.618 ms | 1 - 214 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® SA8255P | 34.618 ms | 1 - 214 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® QCS8450 | 40.494 ms | 0 - 304 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 35.214 ms | 3 - 17 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 27.966 ms | 3 - 3 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 18.144 ms | 3 - 227 MB | NPU
| SalsaNext | QNN_DLC | float | Qualcomm® SA8295P | 45.232 ms | 0 - 225 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 21.849 ms | 1 - 334 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 27.41 ms | 0 - 278 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 18.261 ms | 1 - 1 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Snapdragon® X Elite | 42.112 ms | 1 - 1 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 33.113 ms | 0 - 353 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Snapdragon® 8 Gen 1 Mobile | 62.135 ms | 1 - 403 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 40.875 ms | 1 - 10 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 43.495 ms | 1 - 3 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® SA8775P | 42.425 ms | 2 - 264 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® SA8650P | 42.425 ms | 2 - 264 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® SA8255P | 42.425 ms | 2 - 264 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® QCS8450 | 62.135 ms | 1 - 403 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 42.683 ms | 3 - 12 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 42.112 ms | 1 - 1 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 27.41 ms | 0 - 278 MB | NPU
| SalsaNext | QNN_DLC | w8a16 | Qualcomm® SA8295P | 58.393 ms | 1 - 308 MB | NPU
| SalsaNext | TFLITE | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 13.259 ms | 10 - 245 MB | NPU
| SalsaNext | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 18.222 ms | 10 - 232 MB | NPU
| SalsaNext | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 20.84 ms | 0 - 298 MB | NPU
| SalsaNext | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 39.942 ms | 10 - 320 MB | NPU
| SalsaNext | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 38.131 ms | 10 - 40 MB | NPU
| SalsaNext | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 28.61 ms | 10 - 13 MB | NPU
| SalsaNext | TFLITE | float | Qualcomm® SA8650P | 231.143 ms | 10 - 39 MB | GPU
| SalsaNext | TFLITE | float | Qualcomm® SA8255P | 231.143 ms | 10 - 39 MB | GPU
| SalsaNext | TFLITE | float | Qualcomm® QCS8450 | 39.942 ms | 10 - 320 MB | NPU
| SalsaNext | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 35.099 ms | 10 - 39 MB | NPU
| SalsaNext | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 18.222 ms | 10 - 232 MB | NPU
| SalsaNext | TFLITE | float | Qualcomm® SA7255P | 113.174 ms | 10 - 224 MB | NPU
| SalsaNext | TFLITE | float | Qualcomm® SA8295P | 45.195 ms | 10 - 236 MB | NPU
## License
* The license for the original implementation of SalsaNext can be found
[here](https://github.com/TiagoCortinhal/SalsaNext/blob/master/LICENSE).
## 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).
|