README
¶
Examples
YOLO Benchmarks
Examples for a variety of YOLO versions have been provided, the following table shows benchmarks of how each version performs with respect to resource requirements running on the RK3588. Note that detection results vary depending on the YOLO version.
The data/palace.jpg image file was used for benchmarking.
| Model | Average Total Time | Inference | Post Processing | Rendering |
|---|---|---|---|---|
| YOLOv5s | 29.6ms | 33.6ms | 1.0ms | 4.4ms |
| YOLOv8s | 44.8ms | 47.5ms | 5.0ms | 4.2ms |
| YOLOv10s | 49.4ms | 57.1ms | 2.9ms | 4.1ms |
| YOLOXs | 42.4ms | 48.7ms | 0.5ms | 5.5ms |
| YOLOv8n-pose | 47.5ms | 54.1ms | 1.0ms | 3.0ms |
| YOLOv5s-seg | 207.5ms | 57.0ms | 107.6ms | 75.9ms |
| YOLOv8s-seg | 232.5ms | 67.9ms | 113.3ms | 73.4ms |
The Inference, Post Processing, and Rendering columns show how processing time is split across the Total Time. These figures are derived from the first run on the benchmark so when totaled together they are slightly higher than the Average Total Time.
The Inference column represents processing on the NPU, Post Processing and Rendering values are performed on the CPU.
 |
|
|---|---|
| YOLOv5s - 21 Objects Detected | YOLOv8s - 22 Objects Detected |
 |
 |
| YOLOv10s - 19 Objects Detected | YOLOXs - 29 Objects Detected |
 |
|
| YOLOv8n-pose - 9 Objects Detected | |
 |
 |
| YOLOv5s-seg - 21 Objects Detected | YOLOv8s-seg - 20 Objects Detected |
Directories
¶
| Path | Synopsis |
|---|---|
|
Example code showing how to perform Automatic License Plate Recognition (ALPR) using a License Plate Detection YOLOv8n and LPRNet model
|
Example code showing how to perform Automatic License Plate Recognition (ALPR) using a License Plate Detection YOLOv8n and LPRNet model |
|
Example code showing how to perform inferencing using a LPRnet model
|
Example code showing how to perform inferencing using a LPRnet model |
|
Example code showing how to perform inferencing using a MobileNetv1 model.
|
Example code showing how to perform inferencing using a MobileNetv1 model. |
|
Running multiple Runtimes in a Pool allows you to take advantage of all three NPU cores to significantly reduce average inferencing time.
|
Running multiple Runtimes in a Pool allows you to take advantage of all three NPU cores to significantly reduce average inferencing time. |
|
Example code showing how to perform OCR on an image using PaddleOCR recognition
|
Example code showing how to perform OCR on an image using PaddleOCR recognition |
|
Example code showing how to perform object detection using a YOLOv10 model.
|
Example code showing how to perform object detection using a YOLOv10 model. |
|
Example code showing how to perform object detection using a YOLOv5 model.
|
Example code showing how to perform object detection using a YOLOv5 model. |
|
Example code showing how to perform object detection using a YOLOv8 model.
|
Example code showing how to perform object detection using a YOLOv8 model. |
|
Example code showing how to perform pose using a YOLOv8 model.
|
Example code showing how to perform pose using a YOLOv8 model. |
Click to show internal directories.
Click to hide internal directories.