Initial commit for AXERA AI accelerators

This commit is contained in:
shizhicheng
2025-10-24 09:03:13 +00:00
parent 4e99ee0c33
commit 7b4eaf2d10
9 changed files with 484 additions and 0 deletions
@@ -47,6 +47,11 @@ Frigate supports multiple different detectors that work on different types of ha
- [Synaptics](#synaptics): synap models can run on Synaptics devices(e.g astra machina) with included NPUs.
**AXERA**
- [AXEngine](#axera): axmodels can run on AXERA AI acceleration.
**For Testing**
- [CPU Detector (not recommended for actual use](#cpu-detector-not-recommended): Use a CPU to run tflite model, this is not recommended and in most cases OpenVINO can be used in CPU mode with better results.
@@ -1099,6 +1104,40 @@ model: # required
labelmap_path: /labelmap/coco-80.txt # required
```
## AXERA
Hardware accelerated object detection is supported on the following SoCs:
- AX650N
- AX8850N
This implementation uses the [AXera Pulsar2 Toolchain](https://huggingface.co/AXERA-TECH/Pulsar2).
See the [installation docs](../frigate/installation.md#axera) for information on configuring the AXEngine hardware.
### Configuration
When configuring the AXEngine detector, you have to specify the model name.
#### yolov5s
A yolov5s model is provided in the container at /axmodels and is used by this detector type by default.
Use the model configuration shown below when using the axengine detector with the default axmodel:
```yaml
detectors: # required
axengine: # required
type: axengine # required
model: # required
path: yolov5s_320 # required
width: 320 # required
height: 320 # required
tensor_format: bgr # required
labelmap_path: /labelmap/coco-80.txt # required
```
## Rockchip platform
Hardware accelerated object detection is supported on the following SoCs:
+14
View File
@@ -110,6 +110,20 @@ Frigate supports multiple different detectors that work on different types of ha
| ssd mobilenet | ~ 25 ms |
| yolov5m | ~ 118 ms |
**Synaptics**
- [Synaptics](#synaptics): synap models can run on Synaptics devices(e.g astra machina) with included NPUs to provide efficient object detection.
:::
### AXERA
- **AXEngine** Default model is **yolov5s_320**
| Name | AXERA AX650N/AX8850N Inference Time |
| ---------------- | ----------------------------------- |
| yolov5s_320 | ~ 1.676 ms |
### Hailo-8
Frigate supports both the Hailo-8 and Hailo-8L AI Acceleration Modules on compatible hardware platforms—including the Raspberry Pi 5 with the PCIe hat from the AI kit. The Hailo detector integration in Frigate automatically identifies your hardware type and selects the appropriate default model when a custom model isn’t provided.
+34
View File
@@ -287,6 +287,40 @@ or add these options to your `docker run` command:
Next, you should configure [hardware object detection](/configuration/object_detectors#synaptics) and [hardware video processing](/configuration/hardware_acceleration_video#synaptics).
### AXERA
AXERA accelerators are available in an M.2 form factor, compatible with both Raspberry Pi and Orange Pi. This form factor has also been successfully tested on x86 platforms, making it a versatile choice for various computing environments.
#### Installation
Using AXERA accelerators requires the installation of the AXCL driver. We provide a convenient Linux script to complete this installation.
Follow these steps for installation:
1. Copy or download [this script](https://github.com/ivanshi1108/assets/releases/download/v0.16.2/user_installation.sh).
2. Ensure it has execution permissions with `sudo chmod +x user_installation.sh`
3. Run the script with `./user_installation.sh`
#### Setup
To set up Frigate, follow the default installation instructions, for example: `ghcr.io/blakeblackshear/frigate:stable`
Next, grant Docker permissions to access your hardware by adding the following lines to your `docker-compose.yml` file:
```yaml
devices:
- /dev/axcl_host
- /dev/ax_mmb_dev
- /dev/msg_userdev
```
If you are using `docker run`, add this option to your command `--device /dev/axcl_host --device /dev/ax_mmb_dev --device /dev/msg_userdev`
#### Configuration
Finally, configure [hardware object detection](/configuration/object_detectors#axera) to complete the setup.
## Docker
Running through Docker with Docker Compose is the recommended install method.