Adversarial Inference for Multi-Sentence Video Description

Abstract

While significant progress has been made in the image captioning task, video description is still in its infancy due to the complex nature of video data. Generating multisentence descriptions for long videos is even more challenging. Among the main issues are the fluency and coherence of the generated descriptions, and their relevance to the video. Recently, reinforcement and adversarial learning based methods have been explored to improve the image captioning models; however, both types of methods suffer from a number of issues, e.g. poor readability and high redundancy for RL and stability issues for GANs. In this work, we instead propose to apply adversarial techniques during inference, designing a discriminator which encourages better multi-sentence video description. In addition, we find that a multi-discriminator "hybrid" design, where each discriminator targets one aspect of a description, leads to the best results. Specifically, we decouple the discriminator to evaluate on three criteria: 1) visual relevance to the video, 2) language diversity and fluency, and 3) coherence across sentences. Our approach results in more accurate, diverse, and coherent multi-sentence video descriptions, as shown by automatic as well as human evaluation on the popular ActivityNet Captions dataset.

Applying Deep-Learning-Based Computer Vision to Wireless Communications: Methodologies,Opportunities, and Challenges

“…If they are used in DL network directly, the training time will become extremely long, and much computational resources will be occupied due to the large number of layers. So, commonly, researchers apply the pre-trained ResNet on ImageNet dataset to extract visual features from images, and 3D ResNext on Kinetics dataset to extract spatio-temporal features from videos [11]. Then these features are fed to DL network as part of inputs.…”

“…It can be used to replace the 2D LSTM network. Many CV pieces of research have shown that if these techniques can be jointly applied to make full use of the visual data, better results can be obtained [9], [11]. So, a single proper CV technique or an adequate combination of several CV techniques are required to deal with a specific problem in wireless systems.…”

Relational Graph Learning for Grounded Video Description Generation

“…The prevailing video captioning techniques often incorporate the encoder-decoder pipeline inspired by the first successful sequence-to-sequence model S2VT [30]. Benefitting from the rapid development of deep learning, video captioning models have achieved remarkable advances using attention mechanism [27,39,45], memory networks [3,14,21,31], reinforcement learning [13,20,33] and generative adversarial networks [19,42]. Although these encoderCdecoder-based methods have reached impressive performance on automatic metrics, they often neglect how well the generated caption words (e.g., objects) are grounded in the video, making models less explainable and trustworthy.…”

“…Recently, video captioning [10], the task of automatically generating a sequence of natural-language words to describe a video, has drawn increasing attention [18,19,33,51]. However, these models are known to have poor grounding performance, which leads to objects hallucination [23].…”

Applying Deep-Learning-Based Computer Vision to Wireless Communications: Methodologies,Opportunities, and Challenges

“…It can replace the 2D LSTM network. Much CV research has shown that if these techniques are jointly applied to make full use of the visual data, better results can be obtained [9], [11].…”