administrator/frigate
1
0
Fork 0
mirror of https://github.com/blakeblackshear/frigate.git synced 2026-02-08 04:05:26 +03:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

scratched rocm docs from any mention of edgetpu or openvino

Browse Source
This commit is contained in:
Indrek Mandre 2024-02-09 14:57:39 +02:00
parent d0a882408d
commit 755fbbc4f7
1 changed files with 2 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
docs/docs/configuration/object_detectors.md
View File

@ -515,10 +515,8 @@ services:
Download can be triggered also in regular frigate builds using that environment variable. The following files will be available under `/config/model_cache/yolov8/`:
- `yolov8[ns]_320x320.onnx` -- nano (n) and small (s) sized floating point model files usable by the `rocm`, `onnx` and `openvino` detectors that have been trained using the coco dataset (90 classes)
- `yolov8[ns]-oiv7_320x320.onnx` -- floating point model files usable by the `rocm`, `onnx` and `openvino` detectors that have been trained using the google open images v7 dataset (601 classes)
- `yolov8[ns]-320x320_edgetpu.tflite` and `yolov8[ns]-oiv7_320x320_edgetpu.tflite` -- int8 quantized model files usable by the google coral `edgetpu` detector
- `yolov8[ns]_320x320_i8_openvino.xml` and `yolov8[ns]-oiv7_320x320_i8_openvino.xml` -- int8 quantized model files usable by the `openvino` detector
- `yolov8[ns]_320x320.onnx` -- nano (n) and small (s) sized floating point model files usable by the `rocm` and `onnx` detectors that have been trained using the coco dataset (90 classes)
- `yolov8[ns]-oiv7_320x320.onnx` -- floating point model files usable by the `rocm` and `onnx` detectors that have been trained using the google open images v7 dataset (601 classes)
- `labels.txt` and `labels-frigate.txt` -- full and aggregated labels for the coco dataset models
- `labels-oiv7.txt` and `labels-oiv7-frigate.txt` -- labels for the oiv7 dataset models
@ -550,30 +548,6 @@ Other settings available for the rocm detector
- `conserve_cpu: True` -- run ROCm/HIP synchronization in blocking mode saving CPU (at small loss of latency and maximum throughput)
- `auto_override_gfx: True` -- enable or disable automatic gfx driver detection
#### Advanced configuration
One can configure several types of detectors to run in parallel to increase detection capacity. An example configuration running `rocm`, `edgetpu` and `openvino` in parallel:
```yaml
model:
labelmap_path: /config/model_cache/yolov8/labels-oiv7-frigate.txt
model_type: yolov8
detectors:
rocm:
type: rocm
model:
path: /config/model_cache/yolov8/yolov8s-oiv7_320x320.onnx
coral:
type: edgetpu
device: usb
model:
path: /config/model_cache/yolov8/yolov8s-oiv7_320x320_edgetpu.tflite
openvino:
type: openvino
model:
path: /config/model_cache/yolov8/yolov8s-oiv7_320x320_i8_openvino.xml
```
### Expected performance
On an AMD Ryzen 3 5400U with integrated GPU one can expect getting about 120fps detections with yolov8n and 60fps with yolov8s (320x320).