Image captioning means generate descriptive sentences from a query image automatically. It has recently received widespread attention from the computer vision and natural language processing communities as an emerging visual task. Currently, both components have evolved considerably by exploiting object regions, attributes, attention mechanism methods, entity recognition with novelties, and training strategies. However, despite the impressive results, the research has not yet come to a conclusive answer. This survey aims to provide a comprehensive overview of image captioning methods, from technical architectures to benchmark datasets, evaluation metrics, and comparison of state-of-theart methods. In particular, image captioning methods are divided into different categories based on the technique adopted. Representative methods in each class are summarized, and their advantages and limitations are discussed. Moreover, many related state-of-the-art studies were quantitatively compared to determine the recent trends and future directions in image captioning. The ultimate goal of this work is to serve as a tool for understanding the existing literature and highlighting future directions in the area of image captioning for Computer Vision and Natural Language Processing communities may benefit from.

Images can convey intense affective experiences and affect people on an affective level. With the prevalence of online pictures and videos, evaluating emotions from visual content has attracted considerable attention. Affective image recognition aims to classify the emotions conveyed by digital images automatically. The existing studies using manual features or deep networks mainly focus on low-level visual features or high-level semantic representation without considering all factors. To better understand how deep networks are working for affective recognition tasks, we investigate the convolutional features by visualization them in this work. Our research shows that the hierarchical CNN model mainly relies on deep semantic information while ignoring the shallow visual details, which are essential to evoke emotions. To form a more general and discriminative representation, we propose a multi-level hybrid model that learns and integrates the deep semantics and shallow visual representations for sentiment classification. In addition, this study shows that class imbalance would affect performance as the main category of the affective dataset will overwhelm training and degenerate the deep networks. Therefore, a new loss function is introduced to optimize the deep affective model. Experimental results on several affective image recognition datasets show that our model outperforms various existing studies. The source code is publicly available.

No abstract