Hugging Face's logo

qualcomm
/
RangeNet-Plus-Plus

Other
PyTorch
real_time
bu_auto
android
Model card Files
xet
 Community
RangeNet-Plus-Plus
File size: 11,821 Bytes
29623da
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b38f9e8
29623da
 
 
 
 
 
 
 
 
 
 
 
b38f9e8
 
 
 
 
29623da
 
 
 
 
 
b38f9e8
29623da
 
 
 
 
 
b38f9e8
29623da
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b38f9e8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dcf5ed5
b38f9e8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
29623da
 
 
 
 
 
 
 
 
 
 
 
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
---
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).