3-D object retrieval has been a hot research topic in recent years. Within such a field, view-based approaches are attracting increasing attention because of the flexibility of data representation as well as the reported state-of-the-art performance. One of the most important issues related to view-based 3-D object retrieval is how to learn embedding features that are discriminative across classes while being compactly distributed within each class. In this paper, we analyze the difference between the two tasks of classification and retrieval, and propose a novel way to learn a view-pooling feature via a triplet network. In addition, we propose a new loss, named cube loss, which is able to sample a number of triplets equal to the cube of the samples in a batch. With the new loss, both hard-negative and hard-positive pairs can be effectively investigated. The experimental results on the ModelNet benchmark demonstrate that the proposed method achieves superior performance compared to state-of-the-art approaches.

We present a novel regularization technique for learning local feature descriptors based on statistical information extracted from batches of training samples. With the proposed regularization term, we learn a descriptor distribution in Euclidean space that aims at minimizing the overlap between the distributions of positive pairs and that of negative pairs. The proposed method is able to improve the performance of pairwise and triplet losses with various deep convolution network architectures. This improvement is demonstrated through two different types of architectures, able to obtain state-of-the-art results on the reference benchmark for local feature matching.

To allow collaborative robots to work efficiently and effectively with their human partners, one of the critical functions they need is to precisely and robustly recognize human intentions, i.e., what action they will perform next. In this paper, we present a Recurrent Convolutional Neural Networks (RCNN)-based system that is capable of recognizing a human intention much earlier than the intended action takes place. The system consists of two main components, a Deep Convolutional Neural Networks (DCNN) component that extracts spatial patterns of human activities and a Long Short-Term Memory (LSTM) component that extracts temporal patterns of human activities. We demonstrate the power of our proposed system to data of humans manipulating objects. The results show that our system has superior performance than many existing algorithms in term of recognition accuracy. Moreover, our system can achieve a quite high intention prediction accuracy (about 80%) provided with only the first 80% of the data sequence.