Existing Visual Question Answering (VQA) methods tend to exploit dataset biases and spurious statistical correlations, instead of producing right answers for the right reasons. To address this issue, recent bias mitigation methods for VQA propose to incorporate visual cues (e.g., human attention maps) to better ground the VQA models, showcasing impressive gains. However, we show that the performance improvements are not a result of improved visual grounding, but a regularization effect which prevents over-fitting to linguistic priors. For instance, we find that it is not actually necessary to provide proper, human-based cues; random, insensible cues also result in similar improvements. Based on this observation, we propose a simpler regularization scheme that does not require any external annotations and yet achieves near state-of-the-art performance on VQA-CPv2.
more »
« less
WeaQA: Weak Supervision via Captions for Visual Question Answering
Methodologies for training visual question answering (VQA) models assume the availability of datasets with human-annotated ImageQuestion-Answer (I-Q-A) triplets. This has led to heavy reliance on datasets and a lack of generalization to new types of questions and
scenes. Linguistic priors along with biases and errors due to annotator subjectivity have been
shown to percolate into VQA models trained on such samples. We study whether models
can be trained without any human-annotated Q-A pairs, but only with images and their associated textual descriptions or captions. We present a method to train models with synthetic Q-A pairs generated procedurally from captions. Additionally, we demonstrate the efficacy of spatial-pyramid image patches as a simple but effective alternative to dense and
costly object bounding box annotations used in existing VQA models. Our experiments on
three VQA benchmarks demonstrate the efficacy of this weakly-supervised approach, especially on the VQA-CP challenge, which tests performance under changing linguistic priors.
more »
« less
- Award ID(s):
- 1816039
- NSF-PAR ID:
- 10353899
- Editor(s):
- Zong, Chengqing; Xia, Fei; Li, Wenjie; Navigli, Roberto
- Date Published:
- Journal Name:
- Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021
- Page Range / eLocation ID:
- 3420 to 3435
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Activity Recognition (AR) models perform well with a large number of available training instances. However, in the presence of sensor heterogeneity, sensing biasness and variability of human behaviors and activities and unseen activity classes pose key challenges to adopting and scaling these pre-trained activity recognition models in the new environment. These challenging unseen activities recognition problems are addressed by applying transfer learning techniques that leverage a limited number of annotated samples and utilize the inherent structural patterns among activities within and across the source and target domains. This work proposes a novel AR framework that uses the pre-trained deep autoencoder model and generates features from source and target activity samples. Furthermore, this AR frame-work establishes correlations among activities between the source and target domain by exploiting intra- and inter-class knowledge transfer to mitigate the number of labeled samples and recognize unseen activities in the target domain. We validated the efficacy and effectiveness of our AR framework with three real-world data traces (Daily and Sports, Opportunistic, and Wisdm) that contain 41 users and 26 activities in total. Our AR framework achieves performance gains ≈ 5-6% with 111, 18, and 70 activity samples (20 % annotated samples) for Das, Opp, and Wisdm datasets. In addition, our proposed AR framework requires 56, 8, and 35 fewer activity samples (10% fewer annotated examples) for Das, Opp, and Wisdm, respectively, compared to the state-of-the-art Untran model.more » « less
-
Deep neural networks have achieved great successes on the image captioning task. However, most of the existing models depend heavily on paired image-sentence datasets, which are very expensive to acquire. In this paper, we make the first attempt to train an image captioning model in an unsupervised manner. Instead of relying on manually labeled image-sentence pairs, our proposed model merely requires an image set, a sentence corpus, and an existing visual concept detector. The sentence corpus is used to teach the captioning model how to generate plausible sentences. Meanwhile, the knowledge in the visual concept detector is distilled into the captioning model to guide the model to recognize the visual concepts in an image. In order to further encourage the generated captions to be semantically consistent with the image, the image and caption are projected into a common latent space so that they can reconstruct each other. Given that the existing sentence corpora are mainly designed for linguistic research and are thus with little reference to image contents, we crawl a large-scale image description corpus of two million natural sentences to facilitate the unsupervised image captioning scenario. Experimental results show that our proposed model is able to produce quite promising results without any caption annotations.more » « less
-
Multimodal disinformation, from `deepfakes' to simple edits that deceive, is an important societal problem. Yet at the same time, the vast majority of media edits are harmless -- such as a filtered vacation photo. The difference between this example, and harmful edits that spread disinformation, is one of intent. Recognizing and describing this intent is a major challenge for today's AI systems. We present the task of Edited Media Understanding, requiring models to answer open-ended questions that capture the intent and implications of an image edit. We introduce a dataset for our task, EMU, with 48k question-answer pairs written in rich natural language. We evaluate a wide variety of vision-and-language models for our task, and introduce a new model PELICAN, which builds upon recent progress in pretrained multimodal representations. Our model obtains promising results on our dataset, with humans rating its answers as accurate 40.35% of the time. At the same time, there is still much work to be done -- humans prefer human-annotated captions 93.56% of the time -- and we provide analysis that highlights areas for further progress.more » « less
-
Video Paragraph Captioning aims to generate a multi-sentence description of an untrimmed video with multiple temporal event locations in a coherent storytelling. Following the human perception process, where the scene is effectively understood by decomposing it into visual (e.g. human, animal) and non- visual components (e.g. action, relations) under the mutual influence of vision and language, we first propose a visual- linguistic (VL) feature. In the proposed VL feature, the scene is modeled by three modalities including (i) a global visual environment; (ii) local visual main agents; (iii) linguistic scene elements. We then introduce an autoregressive Transformer- in-Transformer (TinT) to simultaneously capture the semantic coherence of intra- and inter-event contents within a video. Finally, we present a new VL contrastive loss function to guarantee the learnt embedding features are consistent with the captions semantics. Comprehensive experiments and extensive ablation studies on the ActivityNet Captions and YouCookII datasets show that the proposed Visual-Linguistic Transformer- in-Transform (VLTinT) outperforms previous state-of-the-art methods in terms of accuracy and diversity. The source code is made publicly available at: https://github.com/UARK-AICV/ VLTinT.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.18653/v1/2021.findings-acl.302
