Recurrent 3D Attentional Networks for End-to-End Active Object Recognition
This addresses the problem of efficient object recognition in 3D environments for robotics or vision systems, though it is incremental as it builds on existing attention-based models.
The paper tackled multi-view depth-based active object recognition by proposing an end-to-end recurrent 3D attentional network that selects optimal views for fast and accurate recognition, achieving state-of-the-art performance in time efficiency and accuracy.
Active vision is inherently attention-driven: The agent actively selects views to attend in order to fast achieve the vision task while improving its internal representation of the scene being observed. Inspired by the recent success of attention-based models in 2D vision tasks based on single RGB images, we propose to address the multi-view depth-based active object recognition using attention mechanism, through developing an end-to-end recurrent 3D attentional network. The architecture takes advantage of a recurrent neural network (RNN) to store and update an internal representation. Our model, trained with 3D shape datasets, is able to iteratively attend to the best views targeting an object of interest for recognizing it. To realize 3D view selection, we derive a 3D spatial transformer network which is differentiable for training with backpropagation, achieving much faster convergence than the reinforcement learning employed by most existing attention-based models. Experiments show that our method, with only depth input, achieves state-of-the-art next-best-view performance in time efficiency and recognition accuracy.