Shivam Chandhok, Wan-Cyuan Fan, Vered Shwartz, Vineeth N. Balasubramanian, Leonid Sigal. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Ali Salamatian, Amirhossein Abaskohi, Wan-Cyuan Fan, Mir Rayat Imtiaz Hossain, Leonid Sigal, Giuseppe Carenini. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Video Object Segmentation (VOS) has emerged as an increasingly important problem with availability of larger datasets and more complex and realistic settings, which involve long videos with global motion (e.g., in egocentric settings), depicting small objects undergoing both rigid and non-rigid (including state) deformations. While a number of recent approaches have been explored for this task, these data characteristics still present challenges. In this work we propose a novel, clip-based DETR-style encoder-decoder architecture, which focuses on systematically analyzing and addressing aforementioned challenges. Specifically, we propose a novel transformation-aware loss thatfocuses learning on portions of the video where an object undergoes significant deformations - a form of “soft” hard examples mining. Further, we propose a multiplicative time-coded memory, beyond vanilla additive positional encoding, which helps propagate context across long videos. Finally, we incorporate these in our proposed holistic multiscale video transformer for tracking via multi-scale memory matching and decoding to ensure sensitivity and accuracy for long videos and small objects. Our model enables on-line inference with long videos in a windowed fashion, by breaking the video into clips and propagating context among them. We illustrate that short clip length and longer memory with learned time-coding are important design choices for improved performance. Collectively, these technical contributions enable our model to achieve new state-of-the-art (SoTA) performance on two complex egocentric datasets - VISOR [13] and VOST [44], while achieving comparable to SoTA results on the conventional VOS benchmark, DAVIS’ 17 [38]. Detailed ablations validate our design choices and provide insights into the importance of parameter choices and impact on performance.11Code is available at the project page.

Diffusion models (DMs) have shown great potential for high-quality image synthesis. However, when it comes to producing images with complex scenes, how to properly describe both image global structures and object details remains a challenging task. In this paper, we present Frido, a Feature Pyramid Diffusion model performing a multi-scale coarse-to-fine denoising process for image synthesis. Our model decomposes an input image into scale-dependent vector quantized features, followed by a coarse-to-fine gating for producing image output. During the above multi-scale representation learning stage, additional input conditions like text, scene graph, or image layout can be further exploited. Thus, Frido can be also applied for conditional or cross-modality image synthesis. We conduct extensive experiments over various unconditioned and conditional image generation tasks, ranging from text-to-image synthesis, layout-to-image, scene-graph-to-image, to label-to-image. More specifically, we achieved state-of-the-art FID scores on five benchmarks, namely layout-to-image on COCO and OpenImages, scene-graph-to-image on COCO and Visual Genome, and label-to-image on COCO.

We tackle the problem of target-free text-guided image manipulation, which requires one to modify the input reference image based on the given text instruction, while no ground truth target image is observed during training. To address this challenging task, we propose a Cyclic-Manipulation GAN (cManiGAN) in this paper, which is able to realize where and how to edit the image regions of interest. Specifically, the image editor in cManiGAN learns to identify and complete the input image, while cross-modal interpreter and reasoner are deployed to verify the semantic correctness of the output image based on the input instruction. While the former utilizes factual/counterfactual description learning for authenticating the image semantics, the latter predicts the "undo" instruction and provides pixel-level supervision for the training of cManiGAN. With the above operational cycle-consistency, our cManiGAN can be trained in the above weakly supervised setting. We conduct extensive experiments on the datasets of CLEVR and COCO datasets, and the effectiveness and generalizability of our proposed method can be successfully verified. Project page: sites.google.com/view/wancyuanfan/projects/cmanigan.

Object detection over a long-tailed large-scale dataset is practical, challenging, and comprehensively under-explored. Recently proposed methods mainly focus on eliminating the imbalanced classification problem. However, only a few attempts have been made to consider the quality of the predicted bounding boxes. Inspired by the observation of existing Cascade architecture, "detectors with specific IoU thresholds excel at different label frequencies of bounding boxes," this paper first pinpoints the issue in long-tailed distribution. A detector may predict inaccurate bounding boxes on the categories of fewer training data such that the corresponding extracted visual features could further degrade the classification accuracy. Thus, the predicted accuracy of bounding boxes becomes substantially different among categories in the long-tailed distribution. We introduce a Multi-Expert Cascade (MEC) framework that readjusts the weight of each category in the training process via a multi-expert loss. Furthermore, we leverage dynamic ensemble mechanisms at inference time to fully utilize expert detectors and achieve better performance. Extensive experiments on the recent long-tailed large vocabulary object detection dataset show that the proposed MEC framework significantly improves the performance of most widely-used detectors over various backbones on object detection and instance segmentation tasks.

As a key characteristic in audio-visual speech recognition (AVSR), relating linguistic information observed across visual and audio data has been a challenge, benefiting not only audio/visual speech recognition (ASR/VSR) but also for manipulating data within/across modalities. In this paper, we present a feature disentanglement-based framework for jointly addressing the above tasks. By advancing cross-modal mutual learning strategies, our model is able to convert visual or audio-based linguistic features into modality-agnostic representations. Such derived linguistic representations not only allow one to perform ASR, VSR, and AVSR, but also to manipulate audio and visual data output based on the desirable subject identity and linguistic content information. We perform extensive experiments on different recognition and synthesis tasks to show that our model performs favorably against state-of-the-art approaches on each individual task, while ours is a unified solution that is able to jointly tackle the aforementioned audio-visual learning tasks.

In this paper, we address the task of semantics-guided image outpainting, which is to complete an image by generating semantically practical content. Different from most existing image outpainting works, we approach the above task by understanding and completing image semantics at the scene graph level. In particular, we propose a novel network of Scene Graph Transformer (SGT), which is designed to take node and edge features as inputs for modeling the associated structural information. To better understand and process graph-based inputs, our SGT uniquely performs feature attention at both node and edge levels. While the former views edges as relationship regularization, the latter observes the co-occurrence of nodes for guiding the attention process. We demonstrate that, given a partial input image with its layout and scene graph, our SGT can be applied for scene graph expansion and its conversion to a complete layout. Following state-of-the-art layout-to-image conversions works, the task of image outpainting can be completed with sufficient and practical semantics introduced. Extensive experiments are conducted on the datasets of MS-COCO and Visual Genome, which quantitatively and qualitatively confirm the effectiveness of our proposed SGT and outpainting frameworks.

Novel object captioning (NOC) aims to describe images containing objects without observing their ground truth captions during training. Due to the absence of caption annotation, captioning models cannot be directly optimized via sequence-to-sequence training or CIDEr optimization. As a result, we present Paraphrasing-to-Captioning (P2C), a two-stage learning framework for NOC, which would heuristically optimize the output captions via paraphrasing. With P2C, the captioning model first learns paraphrasing from a language model pre-trained on text-only corpus, allowing expansion of the word bank for improving linguistic fluency. To further enforce the output caption sufficiently describing the visual content of the input image, we perform self-paraphrasing for the captioning model with fidelity and adequacy objectives introduced. Since no ground truth captions are available for novel object images during training, our P2C leverages cross-modality (image-text) association modules to ensure the above caption characteristics can be properly preserved. In the experiments, we not only show that our P2C achieves state-of-the-art performances on nocaps and COCO Caption datasets, we also verify the effectiveness and flexibility of our learning framework by replacing language and cross-modality association models for NOC. Implementation details and code are available in the supplementary materials.

When translating text inputs into layouts or images, existing works typically require explicit descriptions of each object in a scene, including their spatial information or the associated relationships. To better exploit the text input, so that implicit objects or relationships can be properly inferred during layout generation, we propose a LayoutTransformer Network (LT-Net) in this paper. Given a scene-graph input, our LT-Net uniquely encodes the semantic features for exploiting their co-occurrences and implicit relationships. This allows one to manipulate conceptually diverse yet plausible layout outputs. Moreover, the decoder of our LT-Net translates the encoded contextual features into bounding boxes with self-supervised relation consistency preserved. By fitting their distributions to Gaussian mixture models, spatially-diverse layouts can be additionally produced by LT-Net. We conduct extensive experiments on the datasets of MS-COCO and Visual Genome, and confirm the effectiveness and plausibility of our LT-Net over recent layout generation models. Codes will be released at LayoutTransformer.