This paper analyzes the predictions of image captioning models with attention mechanisms beyond visualizing the attention itself. We develop variants of layer-wise relevance propagation (LRP) and gradient-based explanation methods, tailored to image captioning models with attention mechanisms. We compare the interpretability of attention heatmaps systematically against the explanations provided by explanation methods such as LRP, Grad-CAM, and Guided Grad-CAM. We show that explanation methods provide simultaneously pixel-wise image explanations (supporting and opposing pixels of the input image) and linguistic explanations (supporting and opposing words of the preceding sequence) for each word in the predicted captions. We demonstrate with extensive experiments that explanation methods 1) can reveal additional evidence used by the model to make decisions compared to attention; 2) correlate to object locations with high precision; 3) are helpful to "debug" the model, e.g. by analyzing the reasons for hallucinated object words. With the observed properties of explanations, we further design an LRP-inference fine-tuning strategy that reduces the issue of object hallucination in image captioning models, and meanwhile, maintains the sentence fluency. We conduct experiments with two widely used attention mechanisms: the adaptive attention mechanism calculated with the additive attention and the multi-head attention mechanism calculated with the scaled dot product.
Hong Kong Baptist University（香港浸会大学）；Zhejiang University（浙江大学）；NetEase Inc.（网易公司）
Image Captioning is a traditional vision-and-language task that aims to generate the language description of an image. Recent studies focus on scaling up the model size and the number of training data, which significantly increase the cost of model training. Different to these heavy-cost models, we introduce a lightweight image captioning framework (I-Tuning), which contains a small number of trainable parameters. We design a novel I-Tuning cross-attention module to connect the non-trainable pre-trained language decoder GPT2 and vision encoder CLIP-ViT. Since most parameters are not required t…
In this paper, we perform a thorough sensitivity analysis on state-of-the-art image captioning approaches using two different architectures: CNN+LSTM and CNN+Transformer. Experiments were carried out using the Flickr8k dataset. The biggest takeaway from the experiments is that fine-tuning the CNN encoder outperforms the baseline and all other experiments carried out for both architectures.
Recurrent Neural Network (RNN) has been widely used to tackle a wide variety of language generation problems and are capable of attaining state-of-the-art (SOTA) performance. However despite its impressive results, the large number of parameters in the RNN model makes deployment to mobile and embedded devices infeasible. Driven by this problem, many works have proposed a number of pruning methods to reduce the sizes of the RNN model. In this work, we propose an end-to-end pruning method for image captioning models equipped with visual attention. Our proposed method is able to achieve sparsity …
Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；University of Science and Technology Beijing（北京科技大学）；Wuhan University（武汉大学）
This report describes our solution for the VATEX Captioning Challenge 2020, which requires generating descriptions for the videos in both English and Chinese languages. We identified three crucial factors that improve the performance, namely: multi-view features, hybrid reward, and diverse ensemble. Based on our method of VATEX 2019 challenge, we achieved significant improvements this year with more advanced model architectures, combination of appearance and motion features, and careful hyper-parameters tuning. Our method achieves very competitive results on both of the Chinese and English vid…
This paper describes our winning entry in the ImageCLEF 2015 image sentence generation task. We improve Google's CNN-LSTM model by introducing concept-based sentence reranking, a data-driven approach which exploits the large amounts of concept-level annotations on Flickr. Different from previous usage of concept detection that is tailored to specific image captioning models, the propose approach reranks predicted sentences in terms of their matches with detected concepts, essentially treating the underlying model as a black box. This property makes the approach applicable to a number of existi…
It is highly desirable yet challenging to generate image captions that can describe novel objects which are unseen in caption-labeled training data, a capability that is evaluated in the novel object captioning challenge (nocaps). In this challenge, no additional image-caption training data, other thanCOCO Captions, is allowed for model training. Thus, conventional Vision-Language Pre-training (VLP) methods cannot be applied. This paper presents VIsual VOcabulary pretraining (VIVO) that performs pre-training in the absence of caption annotations. By breaking the dependency of paired image-capt…
Re$^3$Cap: Retrieval-Guided Refinement for Image Captioning Enhancement via Reinforcement Learning
Taobao & Tmall Group of Alibaba（阿里巴巴淘宝和天猫集团）；The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）
University of Michigan（密歇根大学）；University of Rochester（罗切斯特大学）
Attention mechanisms have attracted considerable interest in image captioning due to its powerful performance. However, existing methods use only visual content as attention and whether textual context can improve attention in image captioning remains unsolved. To explore this problem, we propose a novel attention mechanism, called \textit{text-conditional attention}, which allows the caption generator to focus on certain image features given previously generated text. To obtain text-related image features for our attention model, we adopt the guiding Long Short-Term Memory (gLSTM) captioning …
CyberAgent, Inc.（赛博艾坚特公司）；RIKEN（日本理化学研究所）；Nara Institute of Science and Technology（奈良先端科学技术大学院大学）
Discriminativeness is a desirable feature of image captions: captions should describe the characteristic details of input images. However, recent high-performing captioning models, which are trained with reinforcement learning (RL), tend to generate overly generic captions despite their high performance in various other criteria. First, we investigate the cause of the unexpectedly low discriminativeness and show that RL has a deeply rooted side effect of limiting the output words to high-frequency words. The limited vocabulary is a severe bottleneck for discriminativeness as it is difficult fo…
PIGEONS: Incremental Long-Video Captioning with Persistent Identity Gallery and Evolving Overarching Narrative Summary
Institute of Automation, Chinese Academy of Sciences (CASIA)（中国科学院自动化研究所）；The Hong Kong University of Science and Technology（香港科技大学）；Peking University（北京大学）；School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）；Nanjing University（南京大学）
Exo2EgoDVC: Dense Video Captioning of Egocentric Procedural Activities Using Web Instructional Videos
The University of Tokyo（东京大学）；OMRON SINIC X Corp.（欧姆龙SINIC X公司）；National Institute of Advanced Industrial Science and Technology (AIST)（日本产业技术综合研究所）；LY Corporation（LY公司）
For video captioning, "pre-training and fine-tuning" has become a de facto paradigm, where ImageNet Pre-training (INP) is usually used to encode the video content, then a task-oriented network is fine-tuned from scratch to cope with caption generation. This paper first investigates the impact of the recently proposed CLIP (Contrastive Language-Image Pre-training) on video captioning. Through the empirical study on INP vs. CLIP, we identify the potential deficiencies of INP and explore the key factors for accurate description generation. The results show that the INP-based model is tricky to ca…
National Research University Higher School of Economics（国立研究型大学高等经济学院）；St. Petersburg Department of Steklov Institute of Mathematics（斯捷克洛夫数学研究所圣彼得堡分部）
In this paper a new formulation of event recognition task is examined: it is required to predict event categories in a gallery of images, for which albums (groups of photos corresponding to a single event) are unknown. We propose the novel two-stage approach. At first, features are extracted in each photo using the pre-trained convolutional neural network. These features are classified individually. The scores of the classifier are used to group sequential photos into several clusters. Finally, the features of photos in each group are aggregated into a single descriptor using neural attention …
CapRL++: Unified Reinforcement Learning with Verifiable Rewards for Dense Image and Video Captioning
Tsinghua University（清华大学）；University of Science and Technology of China（中国科学技术大学）；Microsoft（微软）；Shanghai AI Laboratory（上海人工智能实验室）；Shanghai Innovation Institute（上海创新研究院）；Alibaba Cloud（阿里云）；The Chinese University of Hong Kong（香港中文大学）
CCCaption: Dual-Reward Reinforcement Learning for Complete and Correct Image Captioning
Computer Network Information Center, Chinese Academy of Sciences（中国科学院计算机网络信息中心）；Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences（中国科学院大学杭州高等研究院）；LLM Team, Shopee Pte. Ltd.（Shopee私人有限公司大语言模型团队）
A Benchmark for Multi-Lingual Vision-Language Learning in Remote Sensing Image Captioning
School of Artificial Intelligence, Optics and Electronics (iOPEN), Northwestern Polytechnical University（西北工业大学人工智能、光学与电子学院（iOPEN））；Northwest Institute of Nuclear Technology（西北核技术研究所）
RSICCLLM: A Multimodal Large Language Model for Remote Sensing Image Change Captioning
State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China（人工智能安全国家重点实验室，计算技术研究所，中国科学院，北京，中国）；Great Bay University, Dongguan, China（东莞Great Bay大学，中国）；Harbin Institute of Technology, Shenzhen, China（哈尔滨工业大学深圳学院，中国）；Dongguan Key Laboratory for Intelligence and Information Technology, Dongguan, China（东莞智能与信息科技重点实验室，中国）
