C4Synth generates images from multiple captions to improve image quality.
problem Generating images from a single caption is limited; multiple captions are needed.
method Two deep generative models that ensure 'Cross-Caption Cycle Consistency'.
result Quantitative and qualitative validation on Caltech-UCSD Birds and Oxford-102 Flowers datasets.
Generative model uses captions to generate images, improving semantic understanding.
problem Complex image generation models require large datasets and intricate learning.
method Adapts captioning models to generate images, using learned sentence and frame vectors.
result Images generated from multiple captions better capture semantic meaning.
Enhances image captioning with novel context combination methods.
problem Improving machine learning for image captioning with structured learning and meaningful interpretation.
method Combines Feature Distribution Composition (FDC), Multiple Role Representation Crossover (MRRC) attention layers, and language decoder.
result Significantly improved image captioning performance (35.3%) and established new standards.
Efficient video captioning model captures cross-modal interactions.
problem Capturing frame-level cross-modal interactions in video captioning.
method Proposes High-Order Cross-Modal Attention (HOCA) and Low-Rank HOCA.
result Low-Rank HOCA achieves state-of-the-art performance.
Paper develops a framework for generating coherent image captions using visual features and hierarchical topics.
problem Generating semantically coherent paragraphs to describe image content.
method Plug-and-play hierarchical-topic-guided image paragraph generation framework integrating visual extractor and deep topic model.
result Proposed models can distill interpretable multi-layer semantic topics and generate diverse and coherent captions.
New model improves image captioning's ability to describe unseen concepts.
problem Image captioning models struggle with describing unseen combinations of concepts.
method Proposes a multi-task model combining caption generation and image-sentence ranking, with a decoding mechanism to re-rank captions based on image similarity.
result The model significantly outperforms state-of-the-art models in compositional generalization.
Survey of deep learning methods for image captioning.
problem Generating accurate and complex image descriptions.
method Comprehensive review of deep learning techniques for image captioning.
result Analysis of strengths, limitations, and popular datasets in deep learning image captioning.
Improved video and movie description using multitask learning.
problem Lack of training data and poor generalization in video captioning.
method Multitask learning encoder-decoder framework for video sequences.
result Improved performance on multi-caption and single-caption datasets.
Bayesian approach improves image captioning quality metrics.
problem Improving image captioning quality metrics.
method Bayesian Self-Critical Sequence Training (B-SCST) with Monte Carlo dropout approximate variational inference.
result B-SCST improves CIDEr-D scores on various image captioning datasets.
Model learns image-word associations from captions using contrastive learning.
problem Phrase grounding, associating image regions to caption words.
method Optimizing word-region attention to maximize mutual information, using language model guided word substitutions for negatives.
result Model achieves 76.7% accuracy on Flickr30K Entities benchmark, a 5.7% gain from weak supervision.
Transformer model estimates keywords for better audio captioning.
problem Indeterminacy in word selection for audio events/scenes.
method Transformer-based model with keyword estimation.
result Achieved state-of-the-art performance in AAC.
RNNs and their variants have been widely adopted for image captioning. In RNNs, the production of a caption is driven by a sequence of latent states. Existing captioning models usually represent latent states as vectors, taking this practice for granted. We rethink this choice and study an alternative formulation, name…
System tackles indeterminacies in automated audio captioning.
problem Word selection and sentence length indeterminacies in automated audio captioning.
method Solves caption generation and sub-indeterminacy problems through multi-task learning to estimate keywords and sentence length.
result Model achieved 20.7 SPIDEr score, significantly outperforming baseline.
In this paper we study image captioning as a conditional GAN training, proposing both a context-aware LSTM captioner and co-attentive discriminator, which enforces semantic alignment between images and captions. We empirically focus on the viability of two training methods: Self-critical Sequence Training (SCST) and Gu…
This paper examines to what degree current deep learning architectures for image caption generation capture spatial language. On the basis of the evaluation of examples of generated captions from the literature we argue that systems capture what objects are in the image data but not where these objects are located: the…
The New Yorker publishes a weekly captionless cartoon. More than 5,000 readers submit captions for it. The editors select three of them and ask the readers to pick the funniest one. We describe an experiment that compares a dozen automatic methods for selecting the funniest caption. We show that negative sentiment, hum…
JECL clusters images and captions by jointly learning representations and assignments.
problem Clustering image-caption pairs with limited structured training data.
method Parallel encoders trained with clustering and alignment objectives, minimizing KL divergence and maximizing Jensen-Shannon divergence, with regularizers.
result JECL outperforms single-view and multi-view methods on large image-caption datasets.
Paper proposes method to generate images from text using GANs trained on uncaptioned images.
problem Limited captioned image datasets for text-to-image synthesis.
method Conditional GANs trained on uncaptioned images with an Image Captioning Module.
result Promising preliminary results compared to unconditional GANs.
A novel variational autoencoder is developed to model images, as well as associated labels or captions. The Deep Generative Deconvolutional Network (DGDN) is used as a decoder of the latent image features, and a deep Convolutional Neural Network (CNN) is used as an image encoder; the CNN is used to approximate a distri…
Study improves image-caption retrieval by quantifying feature and posterior uncertainty.
problem Improving reliability in image-caption retrieval tasks with deep learning models.
method Quantified feature and posterior uncertainty for model averaging and reliability measure in image-caption retrieval.
result Consistent improvement in retrieval performance with different datasets and architectures.
Study finds CLIP's caption-based learning outperforms image-only methods under certain conditions.
problem Comparing CLIP's performance with traditional image-only methods in learning transferable representations.
method Controlled comparison of CLIP and image-only methods using a dataset with descriptive captions and specific criteria.
result CLIP's caption-based learning outperforms image-only methods when certain conditions are met, but can be detrimental in others.
Text-to-Image translation has been an active area of research in the recent past. The ability for a network to learn the meaning of a sentence and generate an accurate image that depicts the sentence shows ability of the model to think more like humans. Popular methods on text to image translation make use of Generativ…
Improves text-to-image generation with bidirectional capabilities.
problem Generating realistic images from text descriptions.
method Integrates text and image modalities using MMVR architecture with n-gram cost function and multiple sentences.
result Significant improvement in image quality over existing methods (over 20%).
Improved satellite image captions enhance descriptiveness without large models.
problem Extracting meaningful text from satellite imagery.
method Evaluated seven models on a large benchmark, extended vocabulary, and introduced a novel confusion matrix.
result Reduced model size by 100x without sacrificing accuracy, offering new deployment opportunities.
RL for image captions improved with a language prior.
problem Learning biases and large sample space issues in RL image captioning.
method Added a language prior to constrain the action space.
result RL with the language prior module performs better in readability and speed.
Sequence to sequence learning has recently emerged as a new paradigm in supervised learning. To date, most of its applications focused on only one task and not much work explored this framework for multiple tasks. This paper examines three multi-task learning (MTL) settings for sequence to sequence models: (a) the onet…
Planetary exploration missions with Mars rovers are complicated, which generally require elaborated task planning by human experts, from the path to take to the images to capture. NASA has been using this process to acquire over 22 million images from the planet Mars. In order to improve the degree of automation and th…
Area attention allows dynamic area-based attention in memory.
problem Fixed attention granularity limits model performance.
method Area attention dynamically determines area shape and size via learning.
result Area attention improves performance on neural machine translation and image captioning.
Generates realistic faces from detailed textual descriptions.
problem Face generation from fine-grained textual descriptions.
method Conditional GAN model with DC-GAN and GAN-CLS loss, using CelebA dataset with generated captions.
result Promising results in generating diverse face images from fine-grained textual descriptions.
Unified architecture for multi-modal multi-task learning using transformer.
problem Training multiple tasks concurrently with varying modalities.
method Spatio-temporal cache mechanism for multi-modal learning.
result Training multiple tasks together reduces model size by about three times.
LoRA-MCL improves language models by generating diverse sentence continuations.
problem Language models struggle with generating diverse, plausible sentence continuations.
method Low-Rank Adaptation combined with Multiple Choice Learning (MCL) to handle ambiguity.
result LoRA-MCL generates high-diversity and relevant outputs in various tasks.
Pandora hybridizes human and machine methods to explain AI system failures.
problem Understanding and explaining failures in complex AI systems.
method Hybrid human-machine methods and tools for summarizing system malfunction.
result Detailed performance views help in analysis and debugging of AI systems.
The task of associating images and videos with a natural language description has attracted a great amount of attention recently. Rapid progress has been made in terms of both developing novel algorithms and releasing new datasets. Indeed, the state-of-the-art results on some of the standard datasets have been pushed i…
Recent progress on automatic generation of image captions has shown that it is possible to describe the most salient information conveyed by images with accurate and meaningful sentences. In this paper, we propose an image caption system that exploits the parallel structures between images and sentences. In our model, …
Improved representations from multiple video views.
problem Learning reliable representations across unaligned video modalities.
method Correlation-based representation learning on a 4-way parallel multimodal dataset.
result Best models achieve retrieval rates up to 96.9% on open-domain instructional videos.
Deep learning yields great results across many fields, from speech recognition, image classification, to translation. But for each problem, getting a deep model to work well involves research into the architecture and a long period of tuning. We present a single model that yields good results on a number of problems sp…
Generative Score Inference improves uncertainty quantification for multimodal data.
problem Accurate uncertainty quantification in multimodal learning tasks.
method Generative Score Inference (GSI) uses synthetic samples to approximate conditional score distributions.
result GSI achieves state-of-the-art performance in hallucination detection and image captioning uncertainty estimation.
Fine-tunes language models with captions for differential equation solving.
problem Overreliance on function data in operator learning.
method Integrates human knowledge through captions and fine-tunes language models.
result Significantly enhanced performance and reduced function data requirements.
Increasingly many real world tasks involve data in multiple modalities or views. This has motivated the development of many effective algorithms for learning a common latent space to relate multiple domains. However, most existing cross-view learning algorithms assume access to paired data for training. Their applicabi…
Paper tackles model failures producing outputs of undesirable length.
problem Model failures producing outputs of undesirable length.
method Develops a differentiable proxy objective and a verification approach.
result Can produce outputs 50 times longer than input and prove output length below a certain size.
Batch-normalized RHN improves gradient control in recurrent networks.
problem Gradient vanishing or exploding in recurrent networks.
method Batch normalization applied at each recurrence loop in RHN.
result Batch-normalized RHN converges faster and performs better.
Many structured prediction problems (particularly in vision and language domains) are ambiguous, with multiple outputs being correct for an input - e.g. there are many ways of describing an image, multiple ways of translating a sentence; however, exhaustively annotating the applicability of all possible outputs is intr…
Associating image regions with text queries has been recently explored as a new way to bridge visual and linguistic representations. A few pioneering approaches have been proposed based on recurrent neural language models trained generatively (e.g., generating captions), but achieving somewhat limited localization accu…
Pretrained model improves visual dialog performance.
problem Improving performance in visual dialog tasks.
method Pretrained ViLBERT model on vision-language datasets, fine-tuned on VisDial.
result Best model outperforms prior work by more than 1% on NDCG and MRR.
Recurrent Networks are one of the most powerful and promising artificial neural network algorithms to processing the sequential data such as natural languages, sound, time series data. Unlike traditional feed-forward network, Recurrent Network has a inherent feed back loop that allows to store the temporal context info…
Seq-CVAE learns a latent space for each word position to capture sentence intention.
problem Capturing diversity in image captioning models.
method Seq-CVAE learns a sequential latent space for each word position, mimicking future sentence summaries.
result Significantly improves diversity metrics on MSCOCO dataset compared to baselines.
Recurrent Neural Networks (RNNs) are powerful sequence modeling tools. However, when dealing with high dimensional inputs, the training of RNNs becomes computational expensive due to the large number of model parameters. This hinders RNNs from solving many important computer vision tasks, such as Action Recognition in …
Unified framework for generating meteorological time series from text.
problem Lack of large-scale, physically grounded multimodal datasets and architectures ignoring spectral-temporal structure.
method Introduce MeteoCap-3B dataset and MTransformer model.
result State-of-the-art generation quality, accurate cross-modal alignment, strong semantic controllability.