More than 300 million people have been affected by depression all over the world. Due to the medical equipment and knowledge limitations, most of them are not diagnosed at the early stages. Recent work attempts to use social media to detect depression since the patterns of opinions and thoughts expression of the posted text and images, can reflect users' mental state to some extent. In this work, we design a system dubbed SenseMood to demonstrate that the users with depression can be efficiently detected and analyzed by using proposed system. A deep visual-textual multimodal learning approach has been proposed to reveal the psychological state of the users on social networks. The posted images and tweets data from users with/without depression on Twitter have been collected and used for depression detection. CNN-based classifier and Bert are applied to extract the deep features from the pictures and text posted by users respectively. Then visual and textual features are combined to reflect the emotional expression of users. Finally our system classifies the users with depression and normal users through a neural network and the analysis report is generated automatically.

In previous work on joint multiple point sets registration, the multiple point sets are often formulated by a Gaussian mixture model (GMM) and the registration is then cast to a clustering problem, which aims to exploit global relationships on the multiple point sets. However, local relationships on the multiple point sets are ignored in the state-of-the-art joint multiple point sets registration techniques. In this paper, the multiple point sets are assumed to be generated from a GMM. Local features of the multiple point sets, such as shape context, are proposed to infer the membership probabilities of the GMM. The problem of joint multiple point sets registration can be performed by maximum likelihood of the GMM. The parameters of GMM and registration are estimated by an expectation maximization algorithm. Comprehensive experiments demonstrate that our proposed method has better performance than the state-of-the-art methods.

Registration of images from different modalities in the presence of intra-image fluctuation and noise contamination is a challenging task. The accuracy and robustness of the deformable registration largely depend on the definition of appropriate objective function, measuring the similarity between the images. Among them the multi-dimensional modality independent neighbourhood descriptor (MIND) is a promising method, yet its ability is limited by non-uniform bias fields and image noise, etc. Motivated by the fact that Log-Euclidean metric has promising invariance properties such as inversion invariant and similarity invariant, this paper introduces an objective function that embeds Log-Euclidean similarity metric between patches to form a multi-dimensional descriptor. The Gaussian-like penalty function consisting of the log-Euclidean metric between images to be registered is incorporated to better reflect the degree of preserving feature discriminability and structure ordering. Experimental results show the advantages of the proposed method over state-of-the-art techniques both quantitatively and qualitatively.

Since the significant intensity variations existed between different modal images, the deformable registration is still very challenging. In this paper, in order to alleviate the variations deficiency and attain robust alignment, we propose a multi-dimensional tensor based modality independent neighbourhood descriptor (tMIND) to measure the similarity between the images. The tMIND compares the neighboring tensors which consisting of multi-filters induced features. In this work we learn these filters via PCA network (PCANet). We additionally describe the scheme of incorporating these filters into the tMIND. Experimental evaluations demonstrate its promise and effectiveness over the current state-of-the-art approaches.