Visionle is a game of visual word guessing where players have to give names to pictures from the ImageNet-Sketch dataset that are picked at random. Image recognition and word guessing are both parts of the game, which makes it fun and difficult.
We want to correctly guess the word or phrase that best describes a certain picture from the ImageNet-Sketch collection. Players can try as many times as they want to find the right name for the picture. You had to guess which name went with the picture that was shown in each round. In this case, they can use any set of letters that is the same length as or shorter than the real challenge word. There's also a "Looney Mode" that's more relaxed, where players can guess a lot of different things.
The participant keeps guessing based on what they can see in the picture and uses the feedback to figure out the right name. Winners are those who correctly guess the name within the six tries.