Multi-Scale Fusion for Object Representation

• URL: http://arxiv.org/abs/2410.01539v3
• Date: Sun, 09 Nov 2025 11:37:11 GMT
• Title: Multi-Scale Fusion for Object Representation
• Abstract summary: Representing images as object-level feature vectors, rather than pixel-level feature maps, facilitates advanced visual tasks.<n>Existing VAE guidance does not explicitly address that objects can vary in pixel sizes.<n>We propose textitMulti-Scale Fusion (MSF) to enhance VAE guidance for Object-Centric Learning training.
• Score: 29.417271736114454
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Representing images or videos as object-level feature vectors, rather than pixel-level feature maps, facilitates advanced visual tasks. Object-Centric Learning (OCL) primarily achieves this by reconstructing the input under the guidance of Variational Autoencoder (VAE) intermediate representation to drive so-called \textit{slots} to aggregate as much object information as possible. However, existing VAE guidance does not explicitly address that objects can vary in pixel sizes while models typically excel at specific pattern scales. We propose \textit{Multi-Scale Fusion} (MSF) to enhance VAE guidance for OCL training. To ensure objects of all sizes fall within VAE's comfort zone, we adopt the \textit{image pyramid}, which produces intermediate representations at multiple scales; To foster scale-invariance/variance in object super-pixels, we devise \textit{inter}/\textit{intra-scale fusion}, which augments low-quality object super-pixels of one scale with corresponding high-quality super-pixels from another scale. On standard OCL benchmarks, our technique improves mainstream methods, including state-of-the-art diffusion-based ones. The source code is available on https://github.com/Genera1Z/MultiScaleFusion.

Related papers

• MIMONet: Multi-scale Input and Multi-scale Output Network for Salient Object Detection [68.66589458673157]
  Existing methods for saliency detection focus on the application of multi-level features.<n>We propose a Multi-scale Input and Multi-scale Output Network (MIMONet) to improve the recognition ability of a model for objects with different sizes.
  arXiv  Detail & Related papers  (2026-08-26T12:46:22Z)
• MUFASA: A Multi-Layer Framework for Slot Attention [16.325300304610035]
  We introduce MUFASA, a plug-and-play framework for slot attention-based approaches to unsupervised object segmentation.<n>Our model computes slot attention across multiple feature layers of the ViT encoder, fully leveraging their semantic richness.<n>We propose a fusion strategy to aggregate slots obtained on multiple layers into a unified object-centric representation.
  arXiv  Detail & Related papers  (2026-02-07T13:44:56Z)
• PixelRefer: A Unified Framework for Spatio-Temporal Object Referring with Arbitrary Granularity [39.98516860109934]
  PixelRefer is a unified region-level MLLM framework that enables advanced fine-grained understanding over user-specified regions.<n>Our analysis reveals that global visual tokens contribute mainly in early LLM layers, inspiring the design of PixelRefer-Lite.<n>To facilitate fine-grained instruction tuning, we curate PixelRefer-2.2M, a high-quality object-centric instruction dataset.
  arXiv  Detail & Related papers  (2025-10-27T17:59:32Z)
• Revisiting the Integration of Convolution and Attention for Vision Backbone [59.50256661158862]
  Convolutions and multi-head self-attentions (MHSAs) are typically considered alternatives to each other for building vision backbones. We propose in this work to use MSHAs and Convs in parallel textbfat different granularity levels instead. We empirically verify the potential of the proposed integration scheme, named textitGLMix: by offloading the burden of fine-grained features to light-weight Convs, it is sufficient to use MHSAs in a few semantic slots.
  arXiv  Detail & Related papers  (2024-11-21T18:59:08Z)
• Grouped Discrete Representation for Object-Centric Learning [18.44580501357929]
  We propose textitGroup Discrete Representation (GDR) for Object-Centric Learning. GDR decomposes features into attributes via organized channel grouping, and composes these attributes into discrete representation via indexes.
  arXiv  Detail & Related papers  (2024-11-04T17:25:10Z)
• HSVLT: Hierarchical Scale-Aware Vision-Language Transformer for Multi-Label Image Classification [15.129037250680582]
  Tight visual-linguistic interactions play a vital role in improving classification performance. Recent Transformer-based methods have achieved great success in multi-label image classification. We propose a Hierarchical Scale-Aware Vision-Language Transformer (HSVLT) with two appealing designs.
  arXiv  Detail & Related papers  (2024-07-23T07:31:42Z)
• Transcending Fusion: A Multi-Scale Alignment Method for Remote Sensing Image-Text Retrieval [37.775529830620016]
  Remote Sensing Image-Text Retrieval (RSITR) is pivotal for knowledge services and data mining in the remote sensing (RS) domain. Current multi-scale RSITR approaches typically align multi-scale fused image features with text features, but overlook aligning image-text pairs at distinct scales separately. We introduce a novel Multi-Scale Alignment (MSA) method to overcome this limitation.
  arXiv  Detail & Related papers  (2024-05-29T10:19:11Z)
• OV9D: Open-Vocabulary Category-Level 9D Object Pose and Size Estimation [56.028185293563325]
  This paper studies a new open-set problem, the open-vocabulary category-level object pose and size estimation. We first introduce OO3D-9D, a large-scale photorealistic dataset for this task. We then propose a framework built on pre-trained DinoV2 and text-to-image stable diffusion models.
  arXiv  Detail & Related papers  (2024-03-19T03:09:24Z)
• Subobject-level Image Tokenization [60.80949852899857]
  Transformer-based vision models typically tokenize images into fixed-size square patches as input units. Inspired by the subword tokenization widely adopted in language models, we propose an image tokenizer at a subobject level.
  arXiv  Detail & Related papers  (2024-02-22T06:47:44Z)
• Hi-ResNet: Edge Detail Enhancement for High-Resolution Remote Sensing Segmentation [10.919956120261539]
  High-resolution remote sensing (HRS) semantic segmentation extracts key objects from high-resolution coverage areas. objects of the same category within HRS images show significant differences in scale and shape across diverse geographical environments. We propose a High-resolution remote sensing network (Hi-ResNet) with efficient network structure designs.
  arXiv  Detail & Related papers  (2023-05-22T03:58:25Z)
• AutoFocusFormer: Image Segmentation off the Grid [11.257993284839621]
  AutoFocusFormer (AFF) is a local-attention transformer image recognition backbone. We develop a novel point-based local attention block, facilitated by a balanced clustering module. Experiments show that our AutoFocusFormer (AFF) improves significantly over baseline models of similar sizes.
  arXiv  Detail & Related papers  (2023-04-24T19:37:23Z)
• Unicom: Universal and Compact Representation Learning for Image Retrieval [65.96296089560421]
  We cluster the large-scale LAION400M into one million pseudo classes based on the joint textual and visual features extracted by the CLIP model. To alleviate such conflict, we randomly select partial inter-class prototypes to construct the margin-based softmax loss. Our method significantly outperforms state-of-the-art unsupervised and supervised image retrieval approaches on multiple benchmarks.
  arXiv  Detail & Related papers  (2023-04-12T14:25:52Z)
• Scale-MAE: A Scale-Aware Masked Autoencoder for Multiscale Geospatial Representation Learning [55.762840052788945]
  We present Scale-MAE, a pretraining method that explicitly learns relationships between data at different, known scales. We find that tasking the network with reconstructing both low/high frequency images leads to robust multiscale representations for remote sensing imagery.
  arXiv  Detail & Related papers  (2022-12-30T03:15:34Z)
• CoupAlign: Coupling Word-Pixel with Sentence-Mask Alignments for Referring Image Segmentation [104.5033800500497]
  Referring image segmentation aims at localizing all pixels of the visual objects described by a natural language sentence. Previous works learn to straightforwardly align the sentence embedding and pixel-level embedding for highlighting the referred objects. We propose CoupAlign, a simple yet effective multi-level visual-semantic alignment method.
  arXiv  Detail & Related papers  (2022-12-04T08:53:42Z)
• SUNet: Scale-aware Unified Network for Panoptic Segmentation [25.626882426111198]
  We propose two lightweight modules to mitigate the problem of segmenting objects of various scales. We present an end-to-end Scale-aware Unified Network (SUNet) which is more adaptable to multi-scale objects.
  arXiv  Detail & Related papers  (2022-09-07T01:40:41Z)
• Multi-level Second-order Few-shot Learning [111.0648869396828]
  We propose a Multi-level Second-order (MlSo) few-shot learning network for supervised or unsupervised few-shot image classification and few-shot action recognition. We leverage so-called power-normalized second-order base learner streams combined with features that express multiple levels of visual abstraction. We demonstrate respectable results on standard datasets such as Omniglot, mini-ImageNet, tiered-ImageNet, Open MIC, fine-grained datasets such as CUB Birds, Stanford Dogs and Cars, and action recognition datasets such as HMDB51, UCF101, and mini-MIT.
  arXiv  Detail & Related papers  (2022-01-15T19:49:00Z)
• SOIT: Segmenting Objects with Instance-Aware Transformers [16.234574932216855]
  This paper presents an end-to-end instance segmentation framework, termed SOIT, that Segments Objects with Instance-aware Transformers. Inspired by DETR citecarion 2020end, our method views instance segmentation as a direct set prediction problem. Experimental results on the MS COCO dataset demonstrate that SOIT outperforms state-of-the-art instance segmentation approaches significantly.
  arXiv  Detail & Related papers  (2021-12-21T08:23:22Z)
• Improving Semantic Segmentation via Decoupled Body and Edge Supervision [89.57847958016981]
  Existing semantic segmentation approaches either aim to improve the object's inner consistency by modeling the global context, or refine objects detail along their boundaries by multi-scale feature fusion. In this paper, a new paradigm for semantic segmentation is proposed. Our insight is that appealing performance of semantic segmentation requires textitexplicitly modeling the object textitbody and textitedge, which correspond to the high and low frequency of the image. We show that the proposed framework with various baselines or backbone networks leads to better object inner consistency and object boundaries.
  arXiv  Detail & Related papers  (2020-07-20T12:11:22Z)
• Improving Few-shot Learning by Spatially-aware Matching and CrossTransformer [116.46533207849619]
  We study the impact of scale and location mismatch in the few-shot learning scenario. We propose a novel Spatially-aware Matching scheme to effectively perform matching across multiple scales and locations.
  arXiv  Detail & Related papers  (2020-01-06T14:10:20Z)

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.