Finding Multiple Variables Causal Dispositions in Images

In the big data era, one of the fundamental tasks is to explore and discover the causal relationship between variables from observation data and images, which will play a key role in various future applications. This paper reveals in datasets of images that has been emerged the "multiple variables causal direction" of object categories. To start with, it was using an approach to high-dimensional data observational causality inference, and this research can detecting the causal direction between random variables through build a classifier. What is more, this research can validly differentiate the characteristics of objects and their contexts by use causal direction classifier in the static picture. Last but not least, the research certify that there presence a relationship in the causal direction and images' objects or their contexts, also presence visible trace which uncover objects' causal dispositions.