Hi, I want to use FOMO to perform object detection on a wide range of classes. However, it would take a lot of time using dataset like COCO to train it. Is there any pretrained FOMO model available so I can directly deployed to edge device? Thank you.
FOMO already relies on the MobileNetV2 backbone.
Depending on the alpha, we use a different pre-trained mobilenetv2 base models.
Also, FOMO has not been designed to be able to detect a large number of different classes but to be scoped for your business application.
For small general models, you can have a look at the Yolov5n but it is much bigger than FOMO. Both have different purposes.
What edge device are you targeting? Could you share some more context on your use-case?