yolov5x6 detection model 1280x1280 trained on set of 901902-uavs images from a small number of classes (21). current mAP50 of 0.6622. best performance on Kelp and Bird classes
Input images: 1034; transformed images: 18960
18150 Labeled localizations
Input labels: {'Whale': 1, 'Kelp': 14986, 'Foam': 60, 'Bird': 2215, 'Sea_Lion': 33, 'Reflectance': 216, 'Jelly': 94, 'Buoy': 9, 'Otter': 15, 'Boat': 51, 'RIB': 11, 'Person': 4, 'Wave': 6, 'Egregia': 12, 'Seal': 6, 'Shark': 15, 'Mooring': 1, 'Mola': 4, 'Secci_Disc': 4, 'Batray': 1, 'Surfboard': 1};
transformed labels: {'Whale': 4, 'Kelp': 41648, 'Foam': 186, 'Bird': 7251, 'Sea_Lion': 111, 'Reflectance': 552, 'Jelly': 307, 'Buoy': 21, 'Otter': 51, 'Boat': 96, 'RIB': 35, 'Wave': 14, 'Egregia': 38, 'Seal': 22, 'Shark': 57, 'Mooring': 4, 'Mola': 14, 'Secci_Disc': 16, 'Batray': 4, 'Person': 10, 'Surfboard': 4}