arXiv subjects

SouYoung Jin

Publications and source records attributed to SouYoung Jin.

15 recordsLinked to original sources

arXiv2026

StoryTeller: Training-Free Narrative Grounding for Long-Form Audio Description

Long-form audio description (AD) requires more than describing visible actions: it must preserve characters, events, relationships, and story context across scenes so that blind and low-vision (BLV) audiences can follow a film. Modern video-language models (VLMs) are effective on short clips, but they often treat each moment independently, producing descriptions that miss who characters are, why events matter, and how the current scene connects to earlier narrative context. We propose StoryTeller, a training-free framework for story-aware long-form AD. Instead of relying only on local visual cues, StoryTeller maintains a verified narrative memory that carries forward story-relevant information across scenes, enabling later descriptions to remain coherent, grounded, and contextually informative. Given only raw video and a movie title, StoryTeller can optionally retrieve public movie metadata to resolve names and story context, while accepting only facts that are supported by the video through semantic filtering and VLM verification. The method requires no subtitles, scripts, AD transcripts, aligned captions, character banks, precomputed face identities, or task-specific fine-tuning. To evaluate whether generated AD preserves narrative information, we introduce StoryAD-QA, a question-answering benchmark that tests whether a language model can answer story-context questions using only the generated descriptions. Experiments on standard AD benchmarks and diverse long-form videos show that StoryTeller consistently improves narrative coherence, factual grounding, and story comprehension over strong baselines in automatic, QA-based, and human evaluations.

cs.CV↗

arXiv2026

Learning Video Dynamics with Predictive Differentiable Rendering

How to accurately predict a high-fidelity future world? While the visual world is inherently continuous, existing deterministic video prediction models operate in discrete pixel space and are mainly optimized with pixel-wise mean squared error (MSE), which often leads to over-smoothed predictions and a lack of fine-grained visual details. To address these limitations, we propose Predictive Differentiable Rendering (PDR), a novel end-to-end video prediction paradigm that bridges the gap between discrete and continuous representations. Inspired by recent progress in 3D reconstruction with 3D Gaussian Splatting, we introduce PredGS, a lightweight and plug-and-play adapter based on 2D Gaussian representation, which could be seamlessly integrated with existing pixel space predictors, significantly improving spatial detail preservation with negligible computational overhead. Furthermore, we develop predgsplat, a CUDA-accelerated differentiable 2D Gaussian renderer supporting arbitrary channels. Each Gaussian is defined by 5 + C learnable parameters (position, scale, rotation, and C channel amplitudes) and achieves up to 10x faster rendering than the baseline. Optimized by a combined L1 and SSIM loss, PDR overcomes the inherent blurring tendencies of MSE Loss, significantly enhancing the prediction performance. Extensive experiments on diverse real-world benchmarks, including TaxiBJ, WeatherBench, KTH, and Human3.6M, demonstrate that PDR consistently surpasses existing methods, delivering superior detail preservation, visual fidelity, and predictive accuracy.

cs.CV↗

arXiv2026

MoDA: Modulation Adapter for Fine-Grained Visual Grounding in Instructional MLLMs

Multimodal Large Language Models (MLLMs) have achieved remarkable success in instruction-following tasks by integrating pretrained visual encoders with large language models (LLMs). However, existing approaches often struggle with fine-grained visual grounding due to semantic entanglement in visual patch representations, where individual patches blend multiple distinct visual elements, making it difficult for models to focus on instruction-relevant details. To address this challenge, we propose MoDA (Modulation Adapter), a lightweight module that enhances visual grounding through instruction-guided channel-wise modulation. Unlike token-level methods such as Q-Former that perform additive feature selection, MoDA operates at the channel level through multiplicative modulation on already-aligned features, enabling fine-grained control over which embedding dimensions are relevant for each instruction. Following the standard LLaVA training protocol, MoDA applies cross-attention between language instructions and pre-aligned visual features, generating dynamic modulation masks without architectural modifications or additional supervision. We evaluate MoDA across 12 benchmarks spanning visual question answering, vision-centric reasoning, and hallucination detection, including recent 2024 benchmarks (MMVP, CV-Bench, MMStar, RealWorldQA), on three distinct MLLM architectures: LLaVA-1.5, LLaVA-MoRE (2025), and Qwen3-VL (2025). MoDA delivers consistent gains across all three families, with +12.0 points on MMVP for the LLaVA-1.5 family and +4.8 points on ScienceQA for the LLaVA-MoRE family, and +4.9 ScienceQA, +4.1 RealWorldQA, and +3.8 GQA on Qwen3-VL, confirming that the gains generalize beyond CLIP-based encoders with minimal overhead (<1% FLOPs). Code is available at https://github.com/waybarrios/MoDA.

cs.CV↗

arXiv2026

A Foundation Model for Wearable Movement Data in Mental Health Research

Wearable movement data is collected by nearly all commercially available smartwatches and is a valuable resource for mental health research, reflecting fine-grained temporal behavioral trends. Despite its promise, the development of foundation models for health wearable modeling remains limited when compared to clinical image and text analysis. We designed transformers with patch embeddings and used self-supervised masked autoencoder pretraining on minute-level week-long actigraphy (physical activity intensity measurement) sequences to develop and evaluate the Pretrained Actigraphy Transformer (PAT). PAT is an open-source foundation model for wearable movement time series that combines week-long temporal modeling, psychiatric outcome evaluation, and reproducibility on public data. Pretrained on data from 21,538 U.S. participants in a nationally representative cohort from the National Health and Nutrition Examination Survey (NHANES), PAT consistently outperformed non-foundation-model baselines across mental health prediction tasks-including benzodiazepine and SSRI use, depression, and sleep abnormalities. During the benzodiazepine medication usage prediction task, PAT demonstrated the largest improvement over non-foundational deep learning models commonly used for time-series modeling (i.e., 55.6% improvement over the LSTM, 21.4% improvement over the 1-D CNN, 14.8% improvement over the ConvLSTM). Beyond predictive accuracy, PAT provides interpretable attention maps highlighting specific periods of daily activity most important for clinical predictions, offering model transparency and potential clinical insights. The results suggest that PAT offers an easy-to-deploy, adaptable and scalable solution to advance clinical insight from wearable sensor data for researchers and clinicians. GitHub: https://github.com/njacobsonlab/Pretrained-Actigraphy-Transformer/

cs.LG↗

arXiv2026

Unsafe2Safe: Controllable Image Anonymization for Downstream Utility

Large-scale image datasets frequently contain identifiable or sensitive content, raising privacy risks when training models that may memorize and leak such information. We present Unsafe2Safe, a fully automated pipeline that detects privacy-prone images and rewrites only their sensitive regions using multimodally guided diffusion editing. Unsafe2Safe operates in two stages. Stage 1 uses a vision-language model to (i) inspect images for privacy risks, (ii) generate paired private and public captions that respectively include and omit sensitive attributes, and (iii) prompt a large language model to produce structured, identity-neutral edit instructions conditioned on the public caption. Stage 2 employs instruction-driven diffusion editors to apply these dual textual prompts, producing privacy-safe images that preserve global structure and task-relevant semantics while neutralizing private content. To measure anonymization quality, we introduce a unified evaluation suite covering Quality, Cheating, Privacy, and Utility dimensions. Across MS-COCO, Caltech101, and MIT Indoor67, Unsafe2Safe reduces face similarity, text similarity, and demographic predictability by large margins, while maintaining downstream model accuracy comparable to training on raw data. Fine-tuning diffusion editors on our automatically generated triplets (private caption, public caption, edit instruction) further improves both privacy protection and semantic fidelity. Unsafe2Safe provides a scalable, principled solution for constructing large, privacy-safe datasets without sacrificing visual consistency or downstream utility.

cs.CV↗

arXiv2026

Beyond Final Answers: CRYSTAL Benchmark for Transparent Multimodal Reasoning Evaluation

We introduce CRYSTAL (Clear Reasoning via Yielded Steps, Traceability, and Logic), a diagnostic benchmark with 6,372 instances that evaluates multimodal reasoning through verifiable intermediate steps. We propose two complementary metrics: Match F1, which scores step-level precision and recall via semantic similarity matching, and Ordered Match F1, which further penalizes disordered reasoning chains. References are constructed through a Delphi-inspired pipeline in which four independent MLLMs generate trajectories, which are then aggregated via semantic clustering and validated through human quality gates. Evaluation of 20 MLLMs, including commercial frontier systems not used during benchmark construction, reveals systematic failures that are invisible to answer accuracy: universal cherry-picking (precision far exceeds recall), non-monotonic scaling trade-offs, and disordered reasoning in which no competitive model preserves more than 60% of matched steps in the correct order. Beyond evaluation, we propose the Causal Process Reward (CPR), a multiplicative reward that couples answer correctness with step-level alignment, and CPR-Curriculum, which progressively increases reasoning difficulty during training. CPR-Curriculum achieves a 32% improvement in Match F1 via GRPO where additive reward strategies fail, improving reasoning without manual step annotation.

cs.AI↗

arXiv2024

FT2TF: First-Person Statement Text-To-Talking Face Generation

Talking face generation has gained immense popularity in the computer vision community, with various applications including AR, VR, teleconferencing, digital assistants, and avatars. Traditional methods are mainly audio-driven, which have to deal with the inevitable resource-intensive nature of audio storage and processing. To address such a challenge, we propose FT2TF - First-Person Statement Text-To-Talking Face Generation, a novel one-stage end-to-end pipeline for talking face generation driven by first-person statement text. Different from previous work, our model only leverages visual and textual information without any other sources (e.g., audio/landmark/pose) during inference. Extensive experiments are conducted on LRS2 and LRS3 datasets, and results on multi-dimensional evaluation metrics are reported. Both quantitative and qualitative results showcase that FT2TF outperforms existing relevant methods and reaches the state-of-the-art. This achievement highlights our model's capability to bridge first-person statements and dynamic face generation, providing insightful guidance for future work.

cs.CV↗

arXiv2024

Multi-layer Learnable Attention Mask for Multimodal Tasks

While the Self-Attention mechanism in the Transformer model has proven to be effective in many domains, we observe that it is less effective in more diverse settings (e.g. multimodality) due to the varying granularity of each token and the high computational demands of lengthy sequences. To address the challenges, we introduce the Learnable Attention Mask (LAM), strategically designed to globally regulate attention maps and prioritize critical tokens within the sequence. Leveraging the Self-Attention module in a BERT-like transformer network, our approach adeptly captures associations between tokens. The extension of the LAM to a multi-layer version accommodates the varied information aspects embedded at each layer of the Transformer network. Comprehensive experimental validation on various datasets, such as MADv2, QVHighlights, ImageNet 1K, and MSRVTT, demonstrates the efficacy of the LAM, exemplifying its ability to enhance model performance while mitigating redundant computations. This pioneering approach presents a significant advancement in enhancing the understanding of complex scenarios, such as in movie understanding.

cs.CV↗

arXiv2024

LangNav: Language as a Perceptual Representation for Navigation

We explore the use of language as a perceptual representation for vision-and-language navigation (VLN), with a focus on low-data settings. Our approach uses off-the-shelf vision systems for image captioning and object detection to convert an agent's egocentric panoramic view at each time step into natural language descriptions. We then finetune a pretrained language model to select an action, based on the current view and the trajectory history, that would best fulfill the navigation instructions. In contrast to the standard setup which adapts a pretrained language model to work directly with continuous visual features from pretrained vision models, our approach instead uses (discrete) language as the perceptual representation. We explore several use cases of our language-based navigation (LangNav) approach on the R2R VLN benchmark: generating synthetic trajectories from a prompted language model (GPT-4) with which to finetune a smaller language model; domain transfer where we transfer a policy learned on one simulated environment (ALFRED) to another (more realistic) environment (R2R); and combining both vision- and language-based representations for VLN. Our approach is found to improve upon baselines that rely on visual features in settings where only a few expert trajectories (10-100) are available, demonstrating the potential of language as a perceptual representation for navigation.

cs.CV↗

arXiv2023

Learning Human Action Recognition Representations Without Real Humans

Pre-training on massive video datasets has become essential to achieve high action recognition performance on smaller downstream datasets. However, most large-scale video datasets contain images of people and hence are accompanied with issues related to privacy, ethics, and data protection, often preventing them from being publicly shared for reproducible research. Existing work has attempted to alleviate these problems by blurring faces, downsampling videos, or training on synthetic data. On the other hand, analysis on the transferability of privacy-preserving pre-trained models to downstream tasks has been limited. In this work, we study this problem by first asking the question: can we pre-train models for human action recognition with data that does not include real humans? To this end, we present, for the first time, a benchmark that leverages real-world videos with humans removed and synthetic data containing virtual humans to pre-train a model. We then evaluate the transferability of the representation learned on this data to a diverse set of downstream action recognition benchmarks. Furthermore, we propose a novel pre-training strategy, called Privacy-Preserving MAE-Align, to effectively combine synthetic data and human-removed real data. Our approach outperforms previous baselines by up to 5% and closes the performance gap between human and no-human action recognition representations on downstream tasks, for both linear probing and fine-tuning. Our benchmark, code, and models are available at https://github.com/howardzh01/PPMA .

cs.CV↗

arXiv2021

Cross-Modal Discrete Representation Learning

Recent advances in representation learning have demonstrated an ability to represent information from different modalities such as video, text, and audio in a single high-level embedding vector. In this work we present a self-supervised learning framework that is able to learn a representation that captures finer levels of granularity across different modalities such as concepts or events represented by visual objects or spoken words. Our framework relies on a discretized embedding space created via vector quantization that is shared across different modalities. Beyond the shared embedding space, we propose a Cross-Modal Code Matching objective that forces the representations from different views (modalities) to have a similar distribution over the discrete embedding space such that cross-modal objects/actions localization can be performed without direct supervision. In our experiments we show that the proposed discretized multi-modal fine-grained representation (e.g., pixel/word/frame) can complement high-level summary representations (e.g., video/sentence/waveform) for improved performance on cross-modal retrieval tasks. We also observe that the discretized representation uses individual clusters to represent the same semantic concept across modalities.

cs.CV↗

arXiv2021

Spoken Moments: Learning Joint Audio-Visual Representations from Video Descriptions

When people observe events, they are able to abstract key information and build concise summaries of what is happening. These summaries include contextual and semantic information describing the important high-level details (what, where, who and how) of the observed event and exclude background information that is deemed unimportant to the observer. With this in mind, the descriptions people generate for videos of different dynamic events can greatly improve our understanding of the key information of interest in each video. These descriptions can be captured in captions that provide expanded attributes for video labeling (e.g. actions/objects/scenes/sentiment/etc.) while allowing us to gain new insight into what people find important or necessary to summarize specific events. Existing caption datasets for video understanding are either small in scale or restricted to a specific domain. To address this, we present the Spoken Moments (S-MiT) dataset of 500k spoken captions each attributed to a unique short video depicting a broad range of different events. We collect our descriptions using audio recordings to ensure that they remain as natural and concise as possible while allowing us to scale the size of a large classification dataset. In order to utilize our proposed dataset, we present a novel Adaptive Mean Margin (AMM) approach to contrastive learning and evaluate our models on video/caption retrieval on multiple datasets. We show that our AMM approach consistently improves our results and that models trained on our Spoken Moments dataset generalize better than those trained on other video-caption datasets.

cs.CV↗

arXiv2019

Automatic adaptation of object detectors to new domains using self-training

This work addresses the unsupervised adaptation of an existing object detector to a new target domain. We assume that a large number of unlabeled videos from this domain are readily available. We automatically obtain labels on the target data by using high-confidence detections from the existing detector, augmented with hard (misclassified) examples acquired by exploiting temporal cues using a tracker. These automatically-obtained labels are then used for re-training the original model. A modified knowledge distillation loss is proposed, and we investigate several ways of assigning soft-labels to the training examples from the target domain. Our approach is empirically evaluated on challenging face and pedestrian detection tasks: a face detector trained on WIDER-Face, which consists of high-quality images crawled from the web, is adapted to a large-scale surveillance data set; a pedestrian detector trained on clear, daytime images from the BDD-100K driving data set is adapted to all other scenarios such as rainy, foggy, night-time. Our results demonstrate the usefulness of incorporating hard examples obtained from tracking, the advantage of using soft-labels via distillation loss versus hard-labels, and show promising performance as a simple method for unsupervised domain adaptation of object detectors, with minimal dependence on hyper-parameters.

cs.CV↗

arXiv2018

Unsupervised Hard Example Mining from Videos for Improved Object Detection

Important gains have recently been obtained in object detection by using training objectives that focus on {\em hard negative} examples, i.e., negative examples that are currently rated as positive or ambiguous by the detector. These examples can strongly influence parameters when the network is trained to correct them. Unfortunately, they are often sparse in the training data, and are expensive to obtain. In this work, we show how large numbers of hard negatives can be obtained {\em automatically} by analyzing the output of a trained detector on video sequences. In particular, detections that are {\em isolated in time}, i.e., that have no associated preceding or following detections, are likely to be hard negatives. We describe simple procedures for mining large numbers of such hard negatives (and also hard {\em positives}) from unlabeled video data. Our experiments show that retraining detectors on these automatically obtained examples often significantly improves performance. We present experiments on multiple architectures and multiple data sets, including face detection, pedestrian detection and other object categories.

cs.CV↗

arXiv2017

End-to-end Face Detection and Cast Grouping in Movies Using Erdős-Rényi Clustering

We present an end-to-end system for detecting and clustering faces by identity in full-length movies. Unlike works that start with a predefined set of detected faces, we consider the end-to-end problem of detection and clustering together. We make three separate contributions. First, we combine a state-of-the-art face detector with a generic tracker to extract high quality face tracklets. We then introduce a novel clustering method, motivated by the classic graph theory results of Erdős and Rényi. It is based on the observations that large clusters can be fully connected by joining just a small fraction of their point pairs, while just a single connection between two different people can lead to poor clustering results. This suggests clustering using a verification system with very few false positives but perhaps moderate recall. We introduce a novel verification method, rank-1 counts verification, that has this property, and use it in a link-based clustering scheme. Finally, we define a novel end-to-end detection and clustering evaluation metric allowing us to assess the accuracy of the entire end-to-end system. We present state-of-the-art results on multiple video data sets and also on standard face databases.

cs.CV↗