Who is Mistaken?
This work addresses a novel problem in human action understanding, with potential applications in robotics and healthcare, but it appears incremental as it builds on existing belief modeling concepts.
The paper tackles the problem of identifying when characters in abstract scenes have false beliefs, introducing a dataset of 8-frame stories and a method for representing beliefs. Their approach outperforms simple baselines on tasks predicting who is mistaken and when, with diagnostics showing it learns cues like gaze.
Recognizing when people have false beliefs is crucial for understanding their actions. We introduce the novel problem of identifying when people in abstract scenes have incorrect beliefs. We present a dataset of scenes, each visually depicting an 8-frame story in which a character has a mistaken belief. We then create a representation of characters' beliefs for two tasks in human action understanding: predicting who is mistaken, and when they are mistaken. Experiments suggest that our method for identifying mistaken characters performs better on these tasks than simple baselines. Diagnostics on our model suggest it learns important cues for recognizing mistaken beliefs, such as gaze. We believe models of people's beliefs will have many applications in action understanding, robotics, and healthcare.