SEVERE++: Evaluating Benchmark Sensitivity in Generalization of Video Representation Learning

• URL: http://arxiv.org/abs/2504.05706v1
• Date: Tue, 08 Apr 2025 06:00:28 GMT
• Title: SEVERE++: Evaluating Benchmark Sensitivity in Generalization of Video Representation Learning
• Authors: Fida Mohammad Thoker, Letian Jiang, Chen Zhao, Piyush Bagad, Hazel Doughty, Bernard Ghanem, Cees G. M. Snoek,
• Abstract summary: We present a comprehensive evaluation of modern video self-supervised models.<n>We focus on generalization across four key downstream factors: domain shift, sample efficiency, action granularity, and task diversity.<n>Our analysis shows that, despite architectural advances, transformer-based models remain sensitive to downstream conditions.
• Score: 78.44705665291741
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Continued advances in self-supervised learning have led to significant progress in video representation learning, offering a scalable alternative to supervised approaches by removing the need for manual annotations. Despite strong performance on standard action recognition benchmarks, video self-supervised learning methods are largely evaluated under narrow protocols, typically pretraining on Kinetics-400 and fine-tuning on similar datasets, limiting our understanding of their generalization in real world scenarios. In this work, we present a comprehensive evaluation of modern video self-supervised models, focusing on generalization across four key downstream factors: domain shift, sample efficiency, action granularity, and task diversity. Building on our prior work analyzing benchmark sensitivity in CNN-based contrastive learning, we extend the study to cover state-of-the-art transformer-based video-only and video-text models. Specifically, we benchmark 12 transformer-based methods (7 video-only, 5 video-text) and compare them to 10 CNN-based methods, totaling over 1100 experiments across 8 datasets and 7 downstream tasks. Our analysis shows that, despite architectural advances, transformer-based models remain sensitive to downstream conditions. No method generalizes consistently across all factors, video-only transformers perform better under domain shifts, CNNs outperform for fine-grained tasks, and video-text models often underperform despite large scale pretraining. We also find that recent transformer models do not consistently outperform earlier approaches. Our findings provide a detailed view of the strengths and limitations of current video SSL methods and offer a unified benchmark for evaluating generalization in video representation learning.

Related papers

• A Large-Scale Analysis on Contextual Self-Supervised Video Representation Learning [22.870129496984546]
  We establish a unified benchmark that enables fair comparisons across different methods. We investigate five critical aspects of self-supervised learning in videos: (1) dataset size, (2) model complexity, (3) data distribution, (4) data noise, and (5) feature representations. We propose a novel approach that significantly reduces training data requirements while surpassing state-of-the-art methods that rely on 10% more pretraining data.
  arXiv  Detail & Related papers  (2025-04-08T15:47:58Z)
• Learning from Streaming Video with Orthogonal Gradients [62.51504086522027]
  We address the challenge of representation learning from a continuous stream of video as input, in a self-supervised manner.<n>This differs from the standard approaches to video learning where videos are chopped and shuffled during training in order to create a non-redundant batch.<n>We demonstrate the drop in performance when moving from shuffled to sequential learning on three tasks.
  arXiv  Detail & Related papers  (2025-04-02T17:59:57Z)
• Video Annotator: A framework for efficiently building video classifiers using vision-language models and active learning [0.0]
  Video Annotator (VA) is a framework for annotating, managing, and iterating on video classification datasets. VA allows for a continuous annotation process, seamlessly integrating data collection and model training. VA achieves a median 6.8 point improvement in Average Precision relative to the most competitive baseline.
  arXiv  Detail & Related papers  (2024-02-09T17:19:05Z)
• Pre-training Contextualized World Models with In-the-wild Videos for Reinforcement Learning [54.67880602409801]
  In this paper, we study the problem of pre-training world models with abundant in-the-wild videos for efficient learning of visual control tasks. We introduce Contextualized World Models (ContextWM) that explicitly separate context and dynamics modeling. Our experiments show that in-the-wild video pre-training equipped with ContextWM can significantly improve the sample efficiency of model-based reinforcement learning.
  arXiv  Detail & Related papers  (2023-05-29T14:29:12Z)
• Mitigating Representation Bias in Action Recognition: Algorithms and Benchmarks [76.35271072704384]
  Deep learning models perform poorly when applied to videos with rare scenes or objects. We tackle this problem from two different angles: algorithm and dataset. We show that the debiased representation can generalize better when transferred to other datasets and tasks.
  arXiv  Detail & Related papers  (2022-09-20T00:30:35Z)
• Revisiting Classifier: Transferring Vision-Language Models for Video Recognition [102.93524173258487]
  Transferring knowledge from task-agnostic pre-trained deep models for downstream tasks is an important topic in computer vision research. In this study, we focus on transferring knowledge for video classification tasks. We utilize the well-pretrained language model to generate good semantic target for efficient transferring learning.
  arXiv  Detail & Related papers  (2022-07-04T10:00:47Z)
• How Severe is Benchmark-Sensitivity in Video Self-Supervised Learning? [19.920980847895233]
  We investigate how sensitive video self-supervised learning is to the currently used benchmark convention. Our comprehensive set of over 500 experiments reveals that current benchmarks in video self-supervised learning are not a good indicator of generalization.
  arXiv  Detail & Related papers  (2022-03-27T06:32:55Z)
• Self-supervised Video-centralised Transformer for Video Face Clustering [58.12996668434134]
  This paper presents a novel method for face clustering in videos using a video-centralised transformer. We release the first large-scale egocentric video face clustering dataset named EasyCom-Clustering.
  arXiv  Detail & Related papers  (2022-03-24T16:38:54Z)
• Self-supervised Video Representation Learning Using Inter-intra Contrastive Framework [43.002621928500425]
  We propose a self-supervised method to learn feature representations from videos. Because video representation is important, we extend negative samples by introducing intra-negative samples. We conduct experiments on video retrieval and video recognition tasks using the learned video representation.
  arXiv  Detail & Related papers  (2020-08-06T09:08:14Z)

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.