Hi – first many thanks for the updated in 9.60 particularly the AI-masking tools, but your database is quite clearly limited in relation to certain species/types of animals.
I am currently going through the tens of thousands of images from my last safari, where I shot (took photos of) many african big cats. See examples here AND in the new topic I posted in the photos forum
PL9.60 AI-Subject (animals) clearly cannot recognise Leopards OR Cheetah:
- Leopard - Panthera pardus < Panthera < Pantherinae < Felidae (cats) < Suborder: Feliformia < Carnivora (carnivorans) < MAMMALS OF AFRICA < Mammal
- Cheetah - Acinonyx jubatus < Acinonyx < Felinae < Felidae (cats) < Suborder: Feliformia < Carnivora (carnivorans) < Mammal
The AI-Animals mask has less issues with Serval Cats - Leptailurus serval < Leptailurus < Felinae < Felidae (cats) < Suborder: Feliformia < Carnivora (carnivorans) < Mammal – it was somewhat hit or miss. If one moves the subject detection pointer on to the animal in some cases it still misses. A problem of great natural camoflage.
As shown in the attached one can use an undefined AI-mask and provided the target is in the open a mask can be generated. However, I prefer to build and use Presets that I apply to the thousands of images I shoot during a game drive so I prefer to apply 3 standard masks – Subject (Animal); Background (AI-Background, not AI-animal, not AI-sky); and Sky (AI-sky). I accept that these masks are purely a starting point – but to a large extent with Lions and other animals. They work well. But not with Leopards nor Cheetah, and infrequently with Serval cats. I will test with other species.
Please add this species to your list.
AND are these fora the best place to provide feedback where AI-masking tools in PL “struggles” and users like I are seeking improvements.
