File size: 4,903 Bytes
69571fe 5463117 69571fe a9a8cec ac52f0f a9a8cec 51752dd a9a8cec 9178d58 a9a8cec 9178d58 ac52f0f 9178d58 e3cf4b1 9178d58 ac52f0f |
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 |
---
license: apache-2.0
---
<h1>YOLOv8: Target Detection</h1>
<img src="https://cdn-uploads.huggingface.co/production/uploads/64c1fef5b9d81735a12c3fcc/7OYFgkPTO2Os98lS9Mvil.png" width=800>
YOLO algorithm is the most typical representative of one-stage target detection algorithm.
It is based on deep neural network for object recognition and positioning. It runs very fast and can be used in real-time systems. YOLOv8 is currently the most advanced algorithm of the YOLO series, surpassing the previous YOLO series in terms of accuracy and speed.
The model can be found [here](https://github.com/ultralytics/ultralytics)
## CONTENTS
- [Performance](#performance)
- [Model Conversion](#model-conversion)
- [Inference](#inference)
**Performance**
|Device|SoC|Runtime|Model|Size (pixels)|Inference Time (ms)|Precision|Compute Unit|Model Download|
|:----:|:----:|:----:|:----:|:----:|:----:|:----:|:----:|:----:|
|AidBox QCS6490|QCS6490|QNN|YOLOv8s(cutoff)|640|11.1|INT8|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS6490/cutoff_yolov8s_int8.qnn.serialized.bin)|
|AidBox QCS6490|QCS6490|QNN|YOLOv8s(cutoff)|640|24.8|INT16|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS6490/cutoff_yolov8s_int16.qnn.serialized.bin)|
|AidBox QCS6490|QCS6490|SNPE|YOLOv8s(cutoff)|640|9.6|INT8|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS6490/cutoff_yolov8s_int8_htp_snpe2.dlc)|
|AidBox QCS6490|QCS6490|SNPE|YOLOv8s(cutoff)|640|22.1|INT16|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS6490/cutoff_yolov8s_int16_htp_snpe2.dlc)|
|APLUX QCS8550|QCS8550|QNN|YOLOv8s(cutoff)|640|8.7|INT8|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS8550/cutoff_yolov8s_int8.qnn.serialized.bin)|
|APLUX QCS8550|QCS8550|QNN|YOLOv8s(cutoff)|640|20.3|INT16|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS8550/cutoff_yolov8s_int16.qnn.serialized.bin)|
|APLUX QCS8550|QCS8550|SNPE|YOLOv8s(cutoff)|640|3.8|INT8|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS8550/cutoff_yolov8s_int8_htp_snpe2.dlc)|
|APLUX QCS8550|QCS8550|SNPE|YOLOv8s(cutoff)|640|9.3|INT16|NPU|[model download](https://huggingface.co/aidlux/YOLOv8/blob/main/Models/QCS8550/cutoff_yolov8s_int16_htp_snpe2.dlc)|
|AidBox GS865|QCS8250|SNPE|YOLOv8s(cutoff)|640|35|INT8|NPU|[model download]()|
**Models Conversion**
Demo models converted from [**AIMO(AI Model Optimizier)**](https://aidlux.com/en/product/aimo).
The source model **YOLOv8s.onnx** can be found [here](https://huggingface.co/aplux/YOLOv8/blob/main/yolov8s.onnx).
The demo model conversion step on AIMO can be found blow:
|Device|SoC|Runtime|Model|Size (pixels)|Precision|Compute Unit|AIMO Conversion Steps|
|:----:|:----:|:----:|:----:|:----:|:----:|:----:|:----:|
|AidBox QCS6490|QCS6490|QNN|YOLOv8s(cutoff)|640|INT8|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS6490/aimo_yolov8s_qnn_int8.png)|
|AidBox QCS6490|QCS6490|QNN|YOLOv8s(cutoff)|640|INT16|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS6490/aimo_yolov8s_qnn_int16.png)|
|AidBox QCS6490|QCS6490|SNPE|YOLOv8s(cutoff)|640|INT8|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS6490/aimo_yolov8s_snpe_int8.png)|
|AidBox QCS6490|QCS6490|SNPE|YOLOv8s(cutoff)|640|INT16|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS6490/aimo_yolov8s_snpe_int16.png)|
|APLUX QCS8550|QCS8550|QNN|YOLOv8s(cutoff)|640|INT8|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS8550/aimo_yolov8s_qnn_int8.png)|
|APLUX QCS8550|QCS8550|QNN|YOLOv8s(cutoff)|640|INT16|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS8550/aimo_yolov8s_qnn_int16.png)|
|APLUX QCS8550|QCS8550|SNPE|YOLOv8s(cutoff)|640|INT8|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS8550/aimo_yolov8s_snpe_int8.png)|
|APLUX QCS8550|QCS8550|SNPE|YOLOv8s(cutoff)|640|INT16|NPU|[View Steps](https://huggingface.co/aplux/YOLOv8/blob/main/AIMO/QCS8550/aimo_yolov8s_snpe_int16.png)|
|AidBox GS865|QCS8250|SNPE|YOLOv8s(cutoff)|640|INT8|NPU|[View Steps]()|
## Inference
### Step1: convert model
a. Prepare source model in onnx format. The source model can be found [here](https://huggingface.co/aplux/YOLOv8/blob/main/yolov8s.onnx).
b. Login [AIMO](https://aidlux.com/en/product/aimo) and convert source model to target format. The model conversion step can follow **AIMO Conversion Step** in [Model Conversion Sheet](#model-conversion).
c. After conversion task done, download target model file.
### Step2: install AidLite SDK
The installation guide of AidLite SDK can be found [here](https://huggingface.co/datasets/aplux/AIToolKit/blob/main/AidLite%20SDK%20Development%20Documents.md#installation).
### Step3: run demo program
|