Convert YOLONAS ONNX model to OpenVino for Frigate 0.14+

readme.md

After updating frigate to version 0.14, the previous model yolov8 stopped working. Because the author removed its support. This script will allow you to create a new YOLO NAS model for OpenVino

https://github.com/blakeblackshear/frigate/releases/tag/v0.14.0

Other breaking changes
OpenVINO has been upgraded, and Intel Neural Compute sticks are no longer supported. Support for YOLO-NAS models has been added. Additionally, AUTO mode now maps to GPU internally to avoid some unresolved issues with AUTO. The existing YOLO-X model will not work in this version.

YOLO_NAS_OpenVino - for https://colab.research.google.com/
config - example of Frigate config

coco_80cl.txt

person
bicycle
car
motorcycle
airplane
bus
train
truck
boat
trafficlight
firehydrant
stopsign
parkingmeter
bench
bird
cat
dog
horse
sheep
cow
elephant
bear
zebra
giraffe
backpack
umbrella
handbag
tie
suitcase
frisbee
skis
snowboard
sportsball
kite
baseballbat
baseballglove
skateboard
surfboard
tennisracket
bottle
wineglass
cup
fork
knife
spoon
bowl
banana
apple
sandwich
orange
broccoli
carrot
hotdog
pizza
donut
cake
chair
couch
pottedplant
bed
diningtable
toilet
tv
laptop
mouse
remote
keyboard
cellphone
microwave
oven
toaster
sink
refrigerator
book
clock
vase
scissors
teddybear
hairdrier
toothbrush

config.yaml

####### openvino detector for intel 6th+ #######
detectors:
  ov:
    type: openvino
    device: AUTO
    model:
      path: /config/model/yolo_nas_s/yolo_nas_s.xml
model:
  width: 320
  height: 320
  model_type: yolonas
  input_tensor: nchw
  input_pixel_format: bgr
  labelmap_path: /config/model/yolo_nas_s/coco_80cl.txt

YOLO_NAS_OpenVino.py

"""YOLO_NAS_Pretrained_Export.ipynb
Automatically generated by Colab.
"""

! pip install -q super_gradients==3.7.1

from super_gradients.common.object_names import Models
from super_gradients.conversion import DetectionOutputFormatMode
from super_gradients.training import models

model = models.get(Models.YOLO_NAS_S, pretrained_weights="coco")

# export the model for compatibility with Frigate

model.export("yolo_nas_s.onnx",
             output_predictions_format=DetectionOutputFormatMode.FLAT_FORMAT,
             max_predictions_per_image=20,
             confidence_threshold=0.4,
             input_image_shape=(320,320),
            )

!pip install --ignore-installed openvino-dev addict

!python3 /usr/local/bin/mo --input_model yolo_nas_s.onnx --model_name yolo_nas_s --reverse_input_channels --compress_to_fp16 --input_shape [1,3,320,320] --output_dir /content/yolo_nas_s

from google.colab import files
files.download('/content/yolo_nas_s/yolo_nas_s.bin')
files.download('/content/yolo_nas_s/yolo_nas_s.xml')