Inverse Compositional Spatial Transformer Networks
This work addresses alignment and classification challenges in computer vision, offering a novel hybrid approach that is incremental but provides specific gains.
The paper tackled the problem of improving Spatial Transformer Networks (STNs) by connecting them to the Lucas & Kanade algorithm, resulting in Inverse Compositional Spatial Transformer Networks (IC-STNs) that achieve better performance with less model capacity in image alignment and classification tasks.
In this paper, we establish a theoretical connection between the classical Lucas & Kanade (LK) algorithm and the emerging topic of Spatial Transformer Networks (STNs). STNs are of interest to the vision and learning communities due to their natural ability to combine alignment and classification within the same theoretical framework. Inspired by the Inverse Compositional (IC) variant of the LK algorithm, we present Inverse Compositional Spatial Transformer Networks (IC-STNs). We demonstrate that IC-STNs can achieve better performance than conventional STNs with less model capacity; in particular, we show superior performance in pure image alignment tasks as well as joint alignment/classification problems on real-world problems.