Hugging Face's logo

qualcomm
/
OSNet

PyTorch
backbone
real_time
android
video-object-tracking
Model card Files
xet
 Community
OSNet
File size: 12,179 Bytes
c0d592f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52b4833
c0d592f
 
 
 
 
 
 
 
 
 
 
 
52b4833
 
 
 
 
 
c0d592f
 
 
 
 
 
52b4833
c0d592f
 
 
 
 
 
52b4833
c0d592f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52b4833
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c0d592f
 
 
 
 
 
 
 
 
 
 
 
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
---
library_name: pytorch
license: mit
tags:
- backbone
- real_time
- android
pipeline_tag: video-object-tracking

---

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

# OSNet: Optimized for Qualcomm Devices

OSNet is a compact convolutional architecture for person re-identification. It extracts discriminative embeddings that can be compared with cosine similarity for person matching across camera views.

This is based on the implementation of OSNet found [here](https://github.com/KaiyangZhou/deep-person-reid).
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/v0.63.0/src/qai_hub_models/models/osnet) 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 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/osnet/releases/v0.63.0/osnet-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.50, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/osnet/releases/v0.63.0/osnet-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/osnet/releases/v0.63.0/osnet-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/osnet/releases/v0.63.0/osnet-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/osnet/releases/v0.63.0/osnet-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/osnet/releases/v0.63.0/osnet-tflite-w8a8.zip)

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


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.63.0/src/qai_hub_models/models/osnet) 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 [OSNet on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.63.0/src/qai_hub_models/models/osnet) for usage instructions.

## Model Details

**Model Type:** Model_use_case.video_object_tracking

**Model Stats:**
- Input resolution: 256x128
- Model checkpoint: osnet_x1_0_market_256x128_amsgrad_ep150_stp60_lr0.0015_b64_fb10_softmax_labelsmooth_flip.pth
- Model size (float): 8.69 MB
- Model size (w8a8): 3.57 MB
- Number of parameters: 2.68M

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| OSNet | ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.459 ms | 0 - 56 MB | NPU
| OSNet | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.487 ms | 0 - 54 MB | NPU
| OSNet | ONNX | float | Snapdragon® X2 Elite | 0.497 ms | 1 - 1 MB | NPU
| OSNet | ONNX | float | Snapdragon® X Elite | 0.931 ms | 0 - 0 MB | NPU
| OSNet | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 0.607 ms | 0 - 71 MB | NPU
| OSNet | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 1.57 ms | 0 - 71 MB | NPU
| OSNet | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 1.404 ms | 0 - 4 MB | NPU
| OSNet | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.899 ms | 0 - 54 MB | NPU
| OSNet | ONNX | float | Qualcomm® QCS8450 | 1.57 ms | 0 - 71 MB | NPU
| OSNet | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 1.394 ms | 0 - 4 MB | NPU
| OSNet | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.931 ms | 0 - 0 MB | NPU
| OSNet | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 0.487 ms | 0 - 54 MB | NPU
| OSNet | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.281 ms | 0 - 64 MB | NPU
| OSNet | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.326 ms | 0 - 61 MB | NPU
| OSNet | ONNX | w8a8 | Snapdragon® X2 Elite | 0.27 ms | 1 - 1 MB | NPU
| OSNet | ONNX | w8a8 | Snapdragon® X Elite | 0.548 ms | 0 - 0 MB | NPU
| OSNet | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.386 ms | 0 - 77 MB | NPU
| OSNet | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.68 ms | 0 - 80 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 1.446 ms | 0 - 3 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 0.633 ms | 0 - 4 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.548 ms | 0 - 5 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® QCS8450 | 0.68 ms | 0 - 80 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.683 ms | 0 - 3 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.548 ms | 0 - 0 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 2.875 ms | 0 - 66 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 0.586 ms | 0 - 63 MB | NPU
| OSNet | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.326 ms | 0 - 61 MB | NPU
| OSNet | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.586 ms | 0 - 63 MB | NPU
| OSNet | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.433 ms | 0 - 51 MB | NPU
| OSNet | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.529 ms | 0 - 48 MB | NPU
| OSNet | QNN_DLC | float | Snapdragon® X2 Elite | 0.633 ms | 0 - 0 MB | NPU
| OSNet | QNN_DLC | float | Snapdragon® X Elite | 1.237 ms | 0 - 0 MB | NPU
| OSNet | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.694 ms | 0 - 62 MB | NPU
| OSNet | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.978 ms | 0 - 65 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 1.438 ms | 0 - 4 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.046 ms | 0 - 2 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® SA8775P | 1.632 ms | 0 - 52 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® SA8650P | 1.632 ms | 0 - 52 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® SA8255P | 1.632 ms | 0 - 52 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® QCS8450 | 1.978 ms | 0 - 65 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 1.503 ms | 2 - 5 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 1.237 ms | 0 - 0 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 0.529 ms | 0 - 48 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® SA7255P | 4.523 ms | 0 - 48 MB | NPU
| OSNet | QNN_DLC | float | Qualcomm® SA8295P | 2.073 ms | 0 - 43 MB | NPU
| OSNet | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.225 ms | 0 - 48 MB | NPU
| OSNet | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.271 ms | 0 - 46 MB | NPU
| OSNet | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.328 ms | 0 - 0 MB | NPU
| OSNet | QNN_DLC | w8a8 | Snapdragon® X Elite | 0.654 ms | 0 - 0 MB | NPU
| OSNet | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.366 ms | 0 - 61 MB | NPU
| OSNet | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.684 ms | 0 - 65 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 1.438 ms | 0 - 2 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 0.534 ms | 0 - 3 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.524 ms | 0 - 2 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® SA8775P | 0.725 ms | 0 - 52 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® SA8650P | 0.725 ms | 0 - 52 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® SA8255P | 0.725 ms | 0 - 52 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 0.684 ms | 0 - 65 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.618 ms | 0 - 2 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.654 ms | 0 - 0 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 2.858 ms | 0 - 47 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 0.534 ms | 0 - 48 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.271 ms | 0 - 46 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® SA7255P | 1.223 ms | 0 - 48 MB | NPU
| OSNet | QNN_DLC | w8a8 | Qualcomm® SA8295P | 0.851 ms | 0 - 47 MB | NPU
| OSNet | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.534 ms | 0 - 48 MB | NPU
| OSNet | TFLITE | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.438 ms | 0 - 49 MB | NPU
| OSNet | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.532 ms | 0 - 48 MB | NPU
| OSNet | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.697 ms | 0 - 63 MB | NPU
| OSNet | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.972 ms | 0 - 66 MB | NPU
| OSNet | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 1.473 ms | 0 - 8 MB | NPU
| OSNet | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.095 ms | 0 - 3 MB | NPU
| OSNet | TFLITE | float | Qualcomm® QCS8450 | 1.972 ms | 0 - 66 MB | NPU
| OSNet | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 1.542 ms | 0 - 8 MB | NPU
| OSNet | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 0.532 ms | 0 - 48 MB | NPU
| OSNet | TFLITE | float | Qualcomm® SA7255P | 13.319 ms | 0 - 23 MB | GPU
| OSNet | TFLITE | float | Qualcomm® SA8295P | 2.114 ms | 0 - 42 MB | NPU
| OSNet | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 0.208 ms | 0 - 46 MB | NPU
| OSNet | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.248 ms | 0 - 43 MB | NPU
| OSNet | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.307 ms | 0 - 57 MB | NPU
| OSNet | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 0.583 ms | 0 - 63 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 2.243 ms | 0 - 12 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 0.475 ms | 0 - 6 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.438 ms | 0 - 2 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® QCS8450 | 0.583 ms | 0 - 63 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.56 ms | 0 - 5 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 3.696 ms | 0 - 44 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.026 ms | 0 - 46 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.248 ms | 0 - 43 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® SA7255P | 1.085 ms | 0 - 45 MB | NPU
| OSNet | TFLITE | w8a8 | Qualcomm® SA8295P | 0.806 ms | 0 - 44 MB | NPU
| OSNet | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.026 ms | 0 - 46 MB | NPU

## License
* The license for the original implementation of OSNet can be found
  [here](https://github.com/KaiyangZhou/deep-person-reid/blob/master/LICENSE).

## References
* [Omni-Scale Feature Learning for Person Re-Identification](https://arxiv.org/abs/1905.00953)
* [Source Model Implementation](https://github.com/KaiyangZhou/deep-person-reid)

## 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).