Viewpoint invariant semantic object and scene categorization with RGB-D sensors

Indoor Place Category Recognition for a Cleaning Robot by Fusing a Probabilistic Approach and Deep Learning

Indoor place category recognition for a cleaning robot is a problem in which a cleaning robot predicts the category of the indoor place using images captured by it. This is similar to scene recognition in computer vision as well as semantic mapping in robotics. Compared with scene recognition, the indoor place category recognition considered in this article differs as follows: 1) the indoor places include typical home objects; 2) a sequence of images instead of an isolated image is provided because the images are captured successively by a cleaning robot; and 3) the camera of the cleaning robot has a different view compared with those of cameras typically used by human beings. Compared with semantic mapping, indoor place category recognition can be considered as a component in semantic SLAM. In this article, a new method based on the combination of a probabilistic approach and deep learning is proposed to address indoor place category recognition for a cleaning robot. Concerning the probabilistic approach, a new place-object fusion method is proposed based on Bayesian inference. For deep learning, the proposed place-object fusion method is trained using a convolutional neural network in an end-to-end framework. Furthermore, a new recurrent neural network, called the Bayesian filtering network (BFN), is proposed to conduct time-domain fusion. Finally, the proposed method is applied to a benchmark dataset and a new dataset developed in this article, and its validity is demonstrated experimentally.

Convolutional Extreme Learning Machines: A Systematic Review

Much work has recently identified the need to combine deep learning with extreme learning in order to strike a performance balance with accuracy, especially in the domain of multimedia applications. When considering this new paradigm—namely, the convolutional extreme learning machine (CELM)—we present a systematic review that investigates alternative deep learning architectures that use the extreme learning machine (ELM) for faster training to solve problems that are based on image analysis. We detail each of the architectures that are found in the literature along with their application scenarios, benchmark datasets, main results, and advantages, and then present the open challenges for CELM. We followed a well-structured methodology and established relevant research questions that guided our findings. Based on 81 primary studies, we found that object recognition is the most common problem that is solved by CELM, and CCN with predefined kernels is the most common CELM architecture proposed in the literature. The results from experiments show that CELM models present good precision, convergence, and computational performance, and they are able to decrease the total processing time that is required by the learning process. The results presented in this systematic review are expected to contribute to the research area of CELM, providing a good starting point for dealing with some of the current problems in the analysis of computer vision based on images.