README.md

---
library_name: pytorch
license: other
tags:
- real_time
- bu_auto
- android
pipeline_tag: other

---

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

# RangeNet-Plus-Plus: Optimized for Qualcomm Devices

RangeNet-Plus-Plus (also stylized as RangeNet++) projects a LiDAR point cloud onto a 5-channel range image (depth, x, y, z, intensity) and applies a DarkNet-53 encoder with a decoder head to predict per-point semantic class labels in real time.

This is based on the implementation of RangeNet-Plus-Plus found [here](https://github.com/PRBonn/lidar-bonnetal).
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/v0.59.0/src/qai_hub_models/models/rangenet_plus_plus) 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 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/rangenet_plus_plus/releases/v0.59.0/rangenet_plus_plus-onnx-float.zip)
| ONNX | w8a16 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/rangenet_plus_plus/releases/v0.59.0/rangenet_plus_plus-onnx-w8a16.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/rangenet_plus_plus/releases/v0.59.0/rangenet_plus_plus-qnn_dlc-float.zip)
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/rangenet_plus_plus/releases/v0.59.0/rangenet_plus_plus-qnn_dlc-w8a16.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/rangenet_plus_plus/releases/v0.59.0/rangenet_plus_plus-tflite-float.zip)

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


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/rangenet_plus_plus) 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 [RangeNet-Plus-Plus on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/rangenet_plus_plus) for usage instructions.

## Model Details

**Model Type:** Model_use_case.driver_assistance

**Model Stats:**
- Model checkpoint: darknet53_rangenet++
- Input resolution: 64x2048
- Input channels: 5
- Number of output classes: 20
- Backbone: DarkNet-53

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| RangeNet-Plus-Plus | ONNX | float | Snapdragon® X2 Elite | 50.249 ms | 35 - 35 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Snapdragon® X Elite | 99.955 ms | 101 - 101 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 76.463 ms | 0 - 455 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 217.203 ms | 1 - 447 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 102.101 ms | 0 - 138 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Qualcomm® QCS8450 | 217.203 ms | 1 - 447 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 152.799 ms | 21 - 26 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 99.955 ms | 101 - 101 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 59.683 ms | 25 - 344 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Snapdragon® 8 Elite Mobile | 59.683 ms | 25 - 344 MB | NPU
| RangeNet-Plus-Plus | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 43.705 ms | 22 - 363 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Snapdragon® X2 Elite | 37.752 ms | 18 - 18 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Snapdragon® X Elite | 65.56 ms | 52 - 52 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 50.743 ms | 0 - 474 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Snapdragon® 8 Gen 1 Mobile | 111.168 ms | 1 - 584 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 422.1 ms | 13 - 17 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 67.594 ms | 0 - 58 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Qualcomm® QCS8450 | 111.168 ms | 1 - 584 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 68.952 ms | 13 - 17 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 65.56 ms | 52 - 52 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 44.656 ms | 13 - 357 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Snapdragon® 8 Elite Mobile | 44.656 ms | 13 - 357 MB | NPU
| RangeNet-Plus-Plus | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 31.707 ms | 12 - 361 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Snapdragon® X2 Elite | 49.228 ms | 3 - 3 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Snapdragon® X Elite | 99.676 ms | 3 - 3 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 76.707 ms | 3 - 457 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 186.021 ms | 2 - 445 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8275 | 591.087 ms | 2 - 332 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 101.404 ms | 3 - 4 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® SA8775P | 151.457 ms | 1 - 332 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® SA8650P | 151.457 ms | 1 - 332 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® SA8255P | 151.457 ms | 1 - 332 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® QCS8450 | 186.021 ms | 2 - 445 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 151.442 ms | 3 - 17 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 99.676 ms | 3 - 3 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 59.104 ms | 0 - 318 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® SA7255P | 591.087 ms | 2 - 332 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Qualcomm® SA8295P | 168.089 ms | 0 - 325 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 59.104 ms | 0 - 318 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 42.935 ms | 3 - 344 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 35.413 ms | 1 - 1 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Snapdragon® X Elite | 64.031 ms | 1 - 1 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 51.51 ms | 0 - 455 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Snapdragon® 8 Gen 1 Mobile | 102.817 ms | 1 - 568 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 415.398 ms | 3 - 11 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8275 | 192.032 ms | 2 - 322 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 64.24 ms | 1 - 4 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® SA8775P | 62.8 ms | 1 - 322 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® SA8650P | 62.8 ms | 1 - 322 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® SA8255P | 62.8 ms | 1 - 322 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® QCS8450 | 102.817 ms | 1 - 568 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 63.197 ms | 1 - 9 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 64.031 ms | 1 - 1 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 1565.631 ms | 1 - 713 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 120.593 ms | 1 - 445 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 43.882 ms | 2 - 340 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® SA7255P | 192.032 ms | 2 - 322 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Qualcomm® SA8295P | 104.143 ms | 2 - 431 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Snapdragon® 8 Elite Mobile | 43.882 ms | 2 - 340 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 31.168 ms | 1 - 344 MB | NPU
| RangeNet-Plus-Plus | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 120.593 ms | 1 - 445 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 80.492 ms | 10 - 516 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 209.919 ms | 10 - 509 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 595.971 ms | 10 - 317 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 104.209 ms | 10 - 186 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® SA8775P | 155.354 ms | 10 - 317 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® SA8650P | 155.354 ms | 10 - 317 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® SA8255P | 155.354 ms | 10 - 317 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® QCS8450 | 209.919 ms | 10 - 509 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 155.53 ms | 10 - 127 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 62.819 ms | 10 - 303 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® SA7255P | 595.971 ms | 10 - 317 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Qualcomm® SA8295P | 173.168 ms | 10 - 312 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Snapdragon® 8 Elite Mobile | 62.819 ms | 10 - 303 MB | NPU
| RangeNet-Plus-Plus | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 43.71 ms | 0 - 319 MB | NPU

## License
* The license for the original implementation of RangeNet-Plus-Plus can be found
  [here](https://github.com/PRBonn/lidar-bonnetal/blob/master/LICENSE).

## References
* [RangeNet++: Fast and Accurate LiDAR Semantic Segmentation](https://ieeexplore.ieee.org/document/8967762)
* [Source Model Implementation](https://github.com/PRBonn/lidar-bonnetal)

## 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).