We achieve a worst-case On-Device inference time of 60 ms and 76.74% top-3 prediction accuracy with a model size of 3.5 MB.
We trained our model with a carefully crafted dataset of 63 emoji classes and evaluated the performance. The paper focuses on improving user experience and providing low latency on edge devices. We also demonstrate the optimal way to fuse features from both modalities. Here image features are extracted from the stroke-based drawing and text from the previously written context. We present SAMNet, a multimodal deep neural network that jointly learns the text and image features. In this paper, we investigate the effectiveness of combining text and drawing as input to the model. Moreover, the model’s memory footprint and latency play an essential role in providing a seamless writing experience to the user. While the user is digitally writing, it is challenging for the model to identify whether the intention is to write text or draw an emoji. We do not leverage the full context by considering only a single input. To solve this problem, the existing solutions consider either the text or only stroke-based drawings to predict the appropriate emojis. While writing on touch-responsive devices, searching for emojis to capture the true intent is cumbersome. They can also view the artificial intelligence's comparisons of their work with other player-given drawings, before either quitting or replaying.In the current era, the mode of communication through mobile devices is becoming more personalized with the evolution of touch-based input methods. Īt the end of a Quick, Draw! game, the player is given their drawings and results for each round.
A round ends either when the artificial intelligence successfully guesses the drawing or the player runs out of time. During each round, the player is given 20 seconds to draw a random prompt selected from the game's database whilst the artificial intelligence attempts to guess the drawing, akin to a game of Pictionary. In a game of Quick, Draw!, there are six rounds. The concepts that it guesses can be simple, like 'foot', or more complicated, like 'animal migration'. The game is similar to Pictionary in that the player only has a limited time to draw (20 seconds). The AI learns from each drawing, improving its ability to guess correctly in the future.
Quick, Draw! is an online game developed by Google that challenges players to draw a picture of an object or idea and then uses a neural network artificial intelligence to guess what the drawings represent. Jonas Jongejan, Henry Rowley, Takashi Kawashima, Jongmin Kim, Ruben Thomson, Nick Fox-Gieg
0 kommentar(er)
