TWLV-I: Analysis and Insights from Holistic Evaluation on Video Foundation Models

• URL: http://arxiv.org/abs/2408.11318v2
• Date: Fri, 23 Aug 2024 03:33:17 GMT
• Title: TWLV-I: Analysis and Insights from Holistic Evaluation on Video Foundation Models
• Authors: Hyeongmin Lee, Jin-Young Kim, Kyungjune Baek, Jihwan Kim, Hyojun Go, Seongsu Ha, Seokjin Han, Jiho Jang, Raehyuk Jung, Daewoo Kim, GeunOh Kim, JongMok Kim, Jongseok Kim, Junwan Kim, Soonwoo Kwon, Jangwon Lee, Seungjoon Park, Minjoon Seo, Jay Suh, Jaehyuk Yi, Aiden Lee,
• Abstract summary: We present a framework for measuring two core capabilities of video comprehension: appearance and motion understanding. We introduce TWLV-I, a new video foundation model that constructs robust visual representations for both motion- and appearance-based videos. Our model shows a 4.6%p improvement compared to V-JEPA (ViT-L) and a 7.7%p improvement compared to UMT (ViT-L)
• Score: 32.6243916760583
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In this work, we discuss evaluating video foundation models in a fair and robust manner. Unlike language or image foundation models, many video foundation models are evaluated with differing parameters (such as sampling rate, number of frames, pretraining steps, etc.), making fair and robust comparisons challenging. Therefore, we present a carefully designed evaluation framework for measuring two core capabilities of video comprehension: appearance and motion understanding. Our findings reveal that existing video foundation models, whether text-supervised like UMT or InternVideo2, or self-supervised like V-JEPA, exhibit limitations in at least one of these capabilities. As an alternative, we introduce TWLV-I, a new video foundation model that constructs robust visual representations for both motion- and appearance-based videos. Based on the average top-1 accuracy of linear probing on five action recognition benchmarks, pretrained only on publicly accessible datasets, our model shows a 4.6%p improvement compared to V-JEPA (ViT-L) and a 7.7%p improvement compared to UMT (ViT-L). Even when compared to much larger models, our model demonstrates a 7.2%p improvement compared to DFN (ViT-H), a 2.7%p improvement compared to V-JEPA (ViT-H) and a 2.8%p improvement compared to InternVideo2 (ViT-g). We provide embedding vectors obtained by TWLV-I from videos of several commonly used video benchmarks, along with evaluation source code that can directly utilize these embeddings. The code is available at https://github.com/twelvelabs-io/video-embeddings-evaluation-framework.

Related papers

• SEVERE++: Evaluating Benchmark Sensitivity in Generalization of Video Representation Learning [78.44705665291741]
  We present a comprehensive evaluation of modern video self-supervised models. We focus on generalization across four key downstream factors: domain shift, sample efficiency, action granularity, and task diversity. Our analysis shows that, despite architectural advances, transformer-based models remain sensitive to downstream conditions.
  arXiv  Detail & Related papers  (2025-04-08T06:00:28Z)
• The 1st Solution for 4th PVUW MeViS Challenge: Unleashing the Potential of Large Multimodal Models for Referring Video Segmentation [31.44879457190659]
  We propose a simple and effective inference optimization method to fully unleash the potential of LMMs in referring video segmentation. Our solution achieved 61.98% J&F on the MeViS test set and ranked 1st place in the 4th PVUW Challenge MeViS Track at CVPR 2025.
  arXiv  Detail & Related papers  (2025-04-07T15:24:54Z)
• VideoSAVi: Self-Aligned Video Language Models without Human Supervision [0.6854849895338531]
  VideoSAVi is a self-training pipeline that enables Video-LLMs to reason over video content without external supervision. VideoSAVi achieves state-of-the-art performance on MVBench (74.0%) and delivers significant improvements. Our model-agnostic approach is computationally efficient, requiring only 32 frames.
  arXiv  Detail & Related papers  (2024-12-01T00:33:05Z)
• VHELM: A Holistic Evaluation of Vision Language Models [75.88987277686914]
  We present the Holistic Evaluation of Vision Language Models (VHELM) VHELM aggregates various datasets to cover one or more of the 9 aspects: visual perception, knowledge, reasoning, bias, fairness, multilinguality, robustness, toxicity, and safety. Our framework is designed to be lightweight and automatic so that evaluation runs are cheap and fast.
  arXiv  Detail & Related papers  (2024-10-09T17:46:34Z)
• VideoEval: Comprehensive Benchmark Suite for Low-Cost Evaluation of Video Foundation Model [22.188795668927586]
  Video Foundation Models (VFMs) have made significant progress recently. Existing benchmarks and evaluation protocols are often limited by relatively poor diversity, high evaluation costs, and saturated performance metrics. We build a comprehensive benchmark suite to address these issues, namely VideoEval.
  arXiv  Detail & Related papers  (2024-07-09T01:49:08Z)
• PLLaVA : Parameter-free LLaVA Extension from Images to Videos for Video Dense Captioning [78.23573511641548]
  Vision-language pre-training has significantly elevated performance across a wide range of image-language applications. Yet, the pre-training process for video-related tasks demands exceptionally large computational and data resources. This paper investigates a straight-forward, highly efficient, and resource-light approach to adapting an existing image-language pre-trained model for video understanding.
  arXiv  Detail & Related papers  (2024-04-25T19:29:55Z)
• ViTamin: Designing Scalable Vision Models in the Vision-Language Era [26.878662961209997]
  Vision Transformers (ViTs) remain the default choice for the image encoder. ViTamin-L significantly outperforms ViT-L by 2.0% ImageNet zero-shot accuracy. ViTamin-XL with only 436M parameters attains 82.9% ImageNet zero-shot accuracy.
  arXiv  Detail & Related papers  (2024-04-02T17:40:29Z)
• Subjective-Aligned Dataset and Metric for Text-to-Video Quality Assessment [54.00254267259069]
  We establish the largest-scale Text-to-Video Quality Assessment DataBase (T2VQA-DB) to date. The dataset is composed of 10,000 videos generated by 9 different T2V models. We propose a novel transformer-based model for subjective-aligned Text-to-Video Quality Assessment (T2VQA)
  arXiv  Detail & Related papers  (2024-03-18T16:52:49Z)
• Revisiting Feature Prediction for Learning Visual Representations from Video [62.08833572467379]
  V-JEPA is a collection of vision models trained solely using a feature prediction objective. The models are trained on 2 million videos collected from public datasets. Our results show that learning by predicting video features leads to versatile visual representations that perform well on both motion and appearance-based tasks.
  arXiv  Detail & Related papers  (2024-02-15T18:59:11Z)
• TAPIR: Tracking Any Point with per-frame Initialization and temporal Refinement [64.11385310305612]
  We present a novel model for Tracking Any Point (TAP) that effectively tracks any queried point on any physical surface throughout a video sequence. Our approach employs two stages: (1) a matching stage, which independently locates a suitable candidate point match for the query point on every other frame, and (2) a refinement stage, which updates both the trajectory and query features based on local correlations. The resulting model surpasses all baseline methods by a significant margin on the TAP-Vid benchmark, as demonstrated by an approximate 20% absolute average Jaccard (AJ) improvement on DAVIS.
  arXiv  Detail & Related papers  (2023-06-14T17:07:51Z)
• VindLU: A Recipe for Effective Video-and-Language Pretraining [83.49216853881595]
  This paper conducts an empirical study demystifying the most important factors in the VidL model design. Using these empirical insights, we then develop a step-by-step recipe, dubbed VindLU, for effective VidL pretraining. Our model trained using our recipe achieves comparable or better than state-of-the-art results on several VidL tasks.
  arXiv  Detail & Related papers  (2022-12-09T18:54:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.