Self-supervised Normality Learning and Divergence Vector-guided Model Merging for Zero-shot Congenital Heart Disease Detection in Fetal Ultrasound Videos

Congenital Heart Disease (CHD) is one of the leading causes of fetal mortality, yet the scarcity of labeled CHD data and strict privacy regulations surrounding fetal ultrasound (US) imaging present significant challenges for the development of deep learning-based models for CHD detection. Centralised collection of large real-world datasets for rare conditions, such as CHD, from large populations requires significant co-ordination and resource. In addition, data governance rules increasingly prevent data sharing between sites. To address these challenges, we introduce, for the first time, a novel privacy-preserving, zero-shot CHD detection framework that formulates CHD detection as a normality modeling problem integrated with model merging. In our framework dubbed Sparse Tube Ultrasound Distillation (STUD), each hospital site first trains a sparse video tube-based self-supervised video anomaly detection (VAD) model on normal fetal heart US clips with self-distillation loss. This enables site-specific models to independently learn the distribution of healthy cases. To aggregate knowledge across the decentralized models while maintaining privacy, we propose a Divergence Vector-Guided Model Merging approach, DivMerge, that combines site-specific models into a single VAD model without data exchange. Our approach preserves domain-agnostic rich spatio-temporal representations, ensuring generalization to unseen CHD cases. We evaluated our approach on real-world fetal US data collected from 5 hospital sites. Our merged model outperformed site-specific models by 23.77% and 30.13% in accuracy and F1-score respectively on external test sets.

Unsupervised Parameter Efficient Source-free Post-pretraining

Following the success in NLP, the best vision models are now in the billion parameter ranges. Adapting these large models to a target distribution has become computationally and economically prohibitive. Addressing this challenge, we introduce UpStep, an Unsupervised Parameter-efficient Source-free post-pretraining approach, designed to efficiently adapt a base model from a source domain to a target domain: i) we design a self-supervised training scheme to adapt a pretrained model on an unlabeled target domain in a setting where source domain data is unavailable. Such source-free setting comes with the risk of catastrophic forgetting, hence, ii) we propose center vector regularization (CVR), a set of auxiliary operations that minimize catastrophic forgetting and additionally reduces the computational cost by skipping backpropagation in 50\% of the training iterations. Finally iii) we perform this adaptation process in a parameter-efficient way by adapting the pretrained model through low-rank adaptation methods, resulting in a fraction of parameters to optimize. We utilize various general backbone architectures, both supervised and unsupervised, trained on Imagenet as our base model and adapt them to a diverse set of eight target domains demonstrating the adaptability and generalizability of our proposed approach.

Text-guided Synthetic Geometric Augmentation for Zero-shot 3D Understanding

Zero-shot recognition models require extensive training data for generalization. However, in zero-shot 3D classification, collecting 3D data and captions is costly and laborintensive, posing a significant barrier compared to 2D vision. Recent advances in generative models have achieved unprecedented realism in synthetic data production, and recent research shows the potential for using generated data as training data. Here, naturally raising the question: Can synthetic 3D data generated by generative models be used as expanding limited 3D datasets? In response, we present a synthetic 3D dataset expansion method, Textguided Geometric Augmentation (TeGA). TeGA is tailored for language-image-3D pretraining, which achieves SoTA in zero-shot 3D classification, and uses a generative textto-3D model to enhance and extend limited 3D datasets. Specifically, we automatically generate text-guided synthetic 3D data and introduce a consistency filtering strategy to discard noisy samples where semantics and geometric shapes do not match with text. In the experiment to double the original dataset size using TeGA, our approach demonstrates improvements over the baselines, achieving zeroshot performance gains of 3.0% on Objaverse-LVIS, 4.6% on ScanObjectNN, and 8.7% on ModelNet40. These results demonstrate that TeGA effectively bridges the 3D data gap, enabling robust zero-shot 3D classification even with limited real training data and paving the way for zero-shot 3D vision application.

Redefining Normal: A Novel Object-Level Approach for Multi-Object Novelty Detection

In the realm of novelty detection, accurately identifying outliers in data without specific class information poses a significant challenge. While current methods excel in single-object scenarios, they struggle with multi-object situations due to their focus on individual objects. Our paper suggests a novel approach: redefining `normal' at the object level in training datasets. Rather than the usual image-level view, we consider the most dominant object in a dataset as the norm, offering a perspective that is more effective for real-world scenarios. Adapting to our object-level definition of `normal', we modify knowledge distillation frameworks, where a student network learns from a pre-trained teacher network. Our first contribution, DeFeND(Dense Feature Fine-tuning on Normal Data), integrates dense feature fine-tuning into the distillation process, allowing the teacher network to focus on object-level features with a self-supervised loss. The second is masked knowledge distillation, where the student network works with partially hidden inputs, honing its ability to deduce and generalize from incomplete data. This approach not only fares well in single-object novelty detection but also considerably surpasses existing methods in multi-object contexts. The implementation is available at: https://github.com/SMSD75/Redefining_Normal_ACCV24/tree/main

Learning to Ground VLMs without Forgetting

Spatial awareness is key to enable embodied multimodal AI systems. Yet, without vast amounts of spatial supervision, current Visual Language Models (VLMs) struggle at this task. In this paper, we introduce LynX, a framework that equips pretrained VLMs with visual grounding ability without forgetting their existing image and language understanding skills. To this end, we propose a Dual Mixture of Experts module that modifies only the decoder layer of the language model, using one frozen Mixture of Experts (MoE) pre-trained on image and language understanding and another learnable MoE for new grounding capabilities. This allows the VLM to retain previously learned knowledge and skills, while acquiring what is missing. To train the model effectively, we generate a high-quality synthetic dataset we call SCouT, which mimics human reasoning in visual grounding. This dataset provides rich supervision signals, describing a step-by-step multimodal reasoning process, thereby simplifying the task of visual grounding. We evaluate LynX on several object detection and visual grounding datasets, demonstrating strong performance in object detection, zero-shot localization and grounded reasoning while maintaining its original image and language understanding capabilities on seven standard benchmark datasets.

TULIP: Token-length Upgraded CLIP

We address the challenge of representing long captions in vision-language models, such as CLIP. By design these models are limited by fixed, absolute positional encodings, restricting inputs to a maximum of 77 tokens and hindering performance on tasks requiring longer descriptions. Although recent work has attempted to overcome this limit, their proposed approaches struggle to model token relationships over longer distances and simply extend to a fixed new token length. Instead, we propose a generalizable method, named TULIP, able to upgrade the token length to any length for CLIP-like models. We do so by improving the architecture with relative position encodings, followed by a training procedure that (i) distills the original CLIP text encoder into an encoder with relative position encodings and (ii) enhances the model for aligning longer captions with images. By effectively encoding captions longer than the default 77 tokens, our model outperforms baselines on cross-modal tasks such as retrieval and text-to-image generation.

TVBench: Redesigning Video-Language Evaluation

Large language models have demonstrated impressive performance when integrated with vision models even enabling video understanding. However, evaluating these video models presents its own unique challenges, for which several benchmarks have been proposed. In this paper, we show that the currently most used video-language benchmarks can be solved without requiring much temporal reasoning. We identified three main issues in existing datasets: (i) static information from single frames is often sufficient to solve the tasks (ii) the text of the questions and candidate answers is overly informative, allowing models to answer correctly without relying on any visual input (iii) world knowledge alone can answer many of the questions, making the benchmarks a test of knowledge replication rather than visual reasoning. In addition, we found that open-ended question-answering benchmarks for video understanding suffer from similar issues while the automatic evaluation process with LLMs is unreliable, making it an unsuitable alternative. As a solution, we propose TVBench, a novel open-source video multiple-choice question-answering benchmark, and demonstrate through extensive evaluations that it requires a high level of temporal understanding. Surprisingly, we find that most recent state-of-the-art video-language models perform similarly to random performance on TVBench, with only Gemini-Pro and Tarsier clearly surpassing this baseline.

Do better language models have crisper vision?

How well do text-only Large Language Models (LLMs) grasp the visual world? As LLMs are increasingly used in computer vision, addressing this question becomes both fundamental and pertinent. However, existing studies have primarily focused on limited scenarios, such as their ability to generate visual content or cluster multimodal data. To this end, we propose the Visual Text Representation Benchmark (ViTeRB) to isolate key properties that make language models well-aligned with the visual world. With this, we identify large-scale decoder-based LLMs as ideal candidates for representing text in vision-centric contexts, counter to the current practice of utilizing text encoders. Building on these findings, we propose ShareLock, an ultra-lightweight CLIP-like model. By leveraging precomputable frozen features from strong vision and language models, ShareLock achieves an impressive 51% accuracy on ImageNet despite utilizing just 563k image-caption pairs. Moreover, training requires only 1 GPU hour (or 10 hours including the precomputation of features) - orders of magnitude less than prior methods. Code will be released.

Self-Masking Networks for Unsupervised Adaptation

With the advent of billion-parameter foundation models, efficient fine-tuning has become increasingly important for the adaptation of models to downstream tasks. However, especially in computer vision, it can be hard to achieve good performance when access to quality labeled data is lacking. In this work, we propose a method adapting pretrained generalist models in a self-supervised manner by learning binary masks. These self-supervised masking networks (SMNs) are up to 79x more efficient to store and significantly improve performance on label-efficient downstream tasks. We validate the usefulness of learning binary masks as a fine-tuning method on 8 datasets and 3 model architectures, and we demonstrate the effectiveness of SMNs in 3 label-efficient settings.

Foundation Model or Finetune? Evaluation of few-shot semantic segmentation for river pollution

Foundation models (FMs) are a popular topic of research in AI. Their ability to generalize to new tasks and datasets without retraining or needing an abundance of data makes them an appealing candidate for applications on specialist datasets. In this work, we compare the performance of FMs to finetuned pre-trained supervised models in the task of semantic segmentation on an entirely new dataset. We see that finetuned models consistently outperform the FMs tested, even in cases were data is scarce. We release the code and dataset for this work on GitHub.