Unable to upload 8000 images for annotation

[Drakshayani-123]

I have around 50000 images and annotation files for oil refinery object detection task. I want segmentation level detection. But annotation file contains bounding box values. I am not finding any automated segmentation in roboflow. And Roboflow is not taking 80000 images even though it is free up to 10000 images.
For best training yolov8 model how diverse my dataset should be?