Computer vision systems are increasingly used by blind individuals to navigate their lives, helping, for example, locate objects such as doors or chairs. Yet these recognition systems do not work for many personal objects a blind user might want to find, such as keys or a special notebook. In response, efforts created personalized recognition systems, where individuals train their phones to identify and locate things, like a coffee mug or white cane, using example images/videos. However, these tools are trained on data from high-resource contexts, not necessarily reflecting India’s material culture. This paper discusses the contribution of the ORBIT-India dataset, which extends these tools to the Indian context, home of the world’s largest blind population. The ORBIT-India dataset comprises 105,243 images from 587 videos, representing 76 unique objects. We use this experience to examine dataset collection practices translated from high- to low-resource settings, providing recommendations to support cross-geography dataset collection.

Media representations of people with disabilities profoundly influence societal perceptions, yet have historically been absent, stereotyped, or inaccurate. As AI-generated visual media becomes increasingly prevalent, there is a critical opportunity to address these misrepresentations. Responding to the lack of collectively negotiated representation standards, this paper presents our human-centric approach to engaging disability communities meaningfully in AI data practices. Over three months, we worked closely with three disability organizations across the Global North and South to develop the Community Library Creator that introduces design scaffolds to support communities in defining ‘good’ representation and curating community-centric AI datasets; laying the foundations for community-specific evaluation metrics and future model adaptations. We contribute qualitative insights into the complexities of community-led data curation; discuss the value and practical challenges of intersecting human insights with AI requirements; and reflect on human-centered AI approaches that empower communities to share their perspectives and actively shape AI data practices.

Few-shot classification (FSC) entails learning novel classes given only a few examples per class after a pre-training (or meta-training) phase on a set of base classes. Recent works have shown that simply fine-tuning a pre-trained Vision Transformer (ViT) on new test classes is a strong approach for FSC. Fine-tuning ViTs, however, is expensive in time, compute and storage. This has motivated the design of parameter efficient fine-tuning (PEFT) methods which fine-tune only a fraction of the Transformer's parameters. While these methods have shown promise, inconsistencies in experimental conditions make it difficult to disentangle their advantage from other experimental factors including the feature extractor architecture, pre-trained initialization and fine-tuning algorithm, amongst others. In our paper, we conduct a large-scale, experimentally consistent, empirical analysis to study PEFTs for few-shot image classification. Through a battery of over 1.8k controlled experiments on large-scale few-shot benchmarks including Meta-Dataset and ORBIT, we uncover novel insights on PEFTs that cast light on their efficacy in fine-tuning ViTs for few-shot classification. Through our controlled empirical study, we have two main findings: (i) Fine-tuning just the LayerNorm parameters (which we call LN-Tune) during few-shot adaptation is an extremely strong baseline across ViTs pre-trained with both self-supervised and supervised objectives, (ii) For self-supervised ViTs, we find that simply learning a set of scaling parameters for each attention matrix (which we call Attn-Scale) along with a domain-residual adapter (DRA) module leads to state-of-the-art performance (while being ~9x more parameter-efficient) on Meta-Dataset. Our empirical findings set strong baselines and call for rethinking the current design of PEFT methods for FSC.

Large multi-modal models (LMMs) hold the potential to usher in a new era of automated visual assistance for people who are blind or low vision (BLV). Yet, these models have not been systematically evaluated on data captured by BLV users. We address this by empirically assessing CLIP, a widely-used LMM likely to underpin many assistive technologies. Testing 25 CLIP variants in a zero-shot classification task, we find that their accuracy is 15 percentage points lower on average for images captured by BLV users than web-crawled images. This disparity stems from CLIP's sensitivities to 1) image content (e.g. not recognizing disability objects as well as other objects); 2) image quality (e.g. not being robust to lighting variation); and 3) text content (e.g. not recognizing objects described by tactile adjectives as well as visual ones). We delve deeper with a textual analysis of three common pre-training datasets: LAION-400M, LAION-2B and DataComp-1B, showing that disability con-tent is rarely mentioned. We then provide three examples that illustrate how the performance disparities extend to three downstream models underpinned by CLIP: OWL-ViT, CLIPSeg and DALL-E2. We find that few-shot learning with as few as 5 images can mitigate CLIP's quality-of-service disparities for BLV users in some scenarios, which we discuss alongside a set of other possible mitigations.

Image-text contrastive models like CLIP have wide applications in zero-shot classification, image-text retrieval, and transfer learning.However, they often struggle on compositional visio-linguistic tasks (e.g., attribute-binding or object-relationships) where their performance is no better than random chance.To address this, we introduce SDS-CLIP, a lightweight and sample-efficient distillation method to enhance CLIP's compositional visio-linguistic reasoning.Our approach fine-tunes CLIP using a distillation objective borrowed from large text-to-image generative models like Stable-Diffusion, which are known for their strong visio-linguistic reasoning abilities.On the challenging Winoground benchmark, SDS-CLIP improves the visio-linguistic performance of various CLIP models by up to 7%, while on the ARO dataset, it boosts performance by up to 3%.This work underscores the potential of well-designed distillation objectives from generative models to enhance contrastive imagetext models with improved visio-linguistic reasoning capabilities.

Understanding the mechanisms of information storage and transfer in Transformer-based models is important for driving model understanding progress. Recent work has studied these mechanisms for Large Language Models (LLMs), revealing insights on how information is stored in a model's parameters and how information flows to and from these parameters in response to specific prompts. However, these studies have not yet been extended to Multi-modal Large Language Models (MLLMs). Given their expanding capabilities and real-world use, we start by studying one aspect of these models -- how MLLMs process information in a factual visual question answering task. We use a constraint-based formulation which views a visual question as having a set of visual or textual constraints that the model's generated answer must satisfy to be correct (e.g. What movie directed by \emph{the director in this photo} has won a \emph{Golden Globe}?). Under this setting, we contribute i) a method that extends causal information tracing from pure language to the multi-modal setting, and ii) \emph{VQA-Constraints}, a test-bed of 9.7K visual questions annotated with constraints. We use these tools to study two open-source MLLMs, LLaVa and multi-modal Phi-2. Our key findings show that these MLLMs rely on MLP and self-attention blocks in much earlier layers for information storage, compared to LLMs whose mid-layer MLPs are more important. We also show that a consistent small subset of visual tokens output by the vision encoder are responsible for transferring information from the image to these causal blocks. We validate these mechanisms by introducing MultEdit a model-editing algorithm that can correct errors and insert new long-tailed information into MLLMs by targeting these causal blocks. We will publicly release our dataset and code.

The opportunity for artificial intelligence, or AI, to enable accessibility is rapidly growing, but widely impactful applications can be challenging to build given the diversity of user need within and across disability communities. Teachable AI systems give users with disabilities a way to leverage the power of AI to personalize applications for their own specific needs, as long as the effort of providing examples is balanced with the benefit of the personalization received. As an example, this paper presents the design and evaluation of Find My Things, an end-to-end application that can be taught by people who are blind or low vision to find their personal things. Through synthesis of the design process, this paper offers design considerations for the teaching loop that is so critical to realizing the power of teachable AI for accessibility.

Semi-supervised semantic segmentation involves assigning pixel-wise labels to unlabeled images at training time. This is useful in a wide range of real-world applications where collecting pixel-wise labels is not feasible in time or cost. Current approaches to semi-supervised semantic segmentation work by predicting pseudo-labels for each pixel from a class-wise probability distribution output by a model. If this predicted probability distribution is incorrect, however, it leads to poor segmentation results which can have knock-on consequences in safety critical systems, like medical images or self-driving cars. It is, therefore, important to understand what a model does not know, which is mainly achieved by uncertainty quantification. Recently, neural processes (NPs) have been explored in semi-supervised image classification, and they have been a computationally efficient and effective method for uncertainty quantification. In this work, we move one step forward by adapting NPs to semi-supervised semantic segmentation, resulting in a new model called NP-SemiSeg. We experimentally evaluated NP-SemiSeg on the public benchmarks PASCAL VOC 2012 and Cityscapes, with different training settings, and the results verify its effectiveness.

Semi-supervised learning (SSL) has been widely explored in recent years, and it is an effective way of leveraging unlabeled data to reduce the reliance on labeled data. In this work, we adjust neural processes (NPs) to the semi-supervised image classification task, resulting in a new method named NP-Match. NP-Match is suited to this task for two reasons. Firstly, NP-Match implicitly compares data points when making predictions, and as a result, the prediction of each unlabeled data point is affected by the labeled data points that are similar to it, which improves the quality of pseudolabels. Secondly, NP-Match is able to estimate uncertainty that can be used as a tool for selecting unlabeled samples with reliable pseudo-labels. Compared with uncertainty-based SSL methods implemented with Monte Carlo (MC) dropout, NP-Match estimates uncertainty with much less computational overhead, which can save time at both the training and the testing phases. We conducted extensive experiments on four public datasets, and NP-Match outperforms state-of-theart (SOTA) results or achieves competitive results on them, which shows the effectiveness of NPMatch and its potential for SSL.

Artificial Intelligence (AI) for accessibility is a rapidly growing area, requiring datasets that are inclusive of the disabled users that assistive technology aims to serve. We offer insights from a multi-disciplinary project that constructed a dataset for teachable object recognition with people who are blind or low vision. Teachable object recognition enables users to teach a model objects that are of interest to them, e.g., their white cane or own sunglasses, by providing example images or videos of objects. In this paper, we make the following contributions: 1) a disability-first procedure to support blind and low vision data collectors to produce good quality data, using video rather than images; 2) a validation and evolution of this procedure through a series of data collection phases and 3) a set of questions to orient researchers involved in creating datasets toward reflecting on the needs of their participant community.

Object recognition has made great advances in the last decade, but predominately still relies on many high-quality training examples per object category. In contrast, learning new objects from only a few examples could enable many impactful applications from robotics to user personalization. Most few-shot learning research, however, has been driven by benchmark datasets that lack the high variation that these applications will face when deployed in the real-world. To close this gap, we present the ORBIT dataset and benchmark, grounded in the real-world application of teachable object recognizers for people who are blind/low-vision. The dataset contains 3,822 videos of 486 objects recorded by people who are blind/low-vision on their mobile phones. The benchmark reflects a realistic, highly challenging recognition problem, providing a rich playground to drive research in robustness to few-shot, high-variation conditions. We set the benchmark’s first state-of-the-art and show there is massive scope for further innovation, holding the potential to impact a broad range of real-world vision applications including tools for the blind/low-vision community. We release the dataset at https://doi.org/10.25383/city.14294597 and benchmark code at https://github.com/microsoft/ORBIT-Dataset.

Meta learning approaches to few-shot classification are computationally efficient at test time, requiring just a few optimization steps or single forward pass to learn a new task, but they remain highly memory-intensive to train. This limitation arises because a task's entire support set, which can contain up to 1000 images, must be processed before an optimization step can be taken. Harnessing the performance gains offered by large images thus requires either parallelizing the meta-learner across multiple GPUs, which may not be available, or trade-offs between task and image size when memory constraints apply. We improve on both options by proposing LITE, a general and memory efficient episodic training scheme that enables meta-training on large tasks composed of large images on a single GPU. We achieve this by observing that the gradients for a task can be decomposed into a sum of gradients over the task's training images. This enables us to perform a forward pass on a task's entire training set but realize significant memory savings by back-propagating only a random subset of these images which we show is an unbiased approximation of the full gradient. We use LITE to train meta-learners and demonstrate new state-of-the-art accuracy on the real-world ORBIT benchmark and 3 of the 4 parts of the challenging VTAB+MD benchmark relative to leading meta-learners. LITE also enables meta-learners to be competitive with transfer learning approaches but at a fraction of the test-time computational cost, thus serving as a counterpoint to the recent narrative that transfer learning is all you need for few-shot classification.

We present FLIPDIAL, a generative model for Visual Dialogue that simultaneously plays the role of both participants in a visually-grounded dialogue. Given context in the form of an image and an associated caption summarising the contents of the image, FLIPDIAL learns both to answer questions and put forward questions, capable of generating entire sequences of dialogue (question-answer pairs) which are diverse and relevant to the image. To do this, FLIPDIAL relies on a simple but surprisingly powerful idea: it uses convolutional neural networks (CNNs) to encode entire dialogues directly, implicitly capturing dialogue context, and conditional VAEs to learn the generative model, FLIPDIAL outperforms the state-of-the-art model in the sequential answering task (1VD) on the VisDial dataset by 5 points in Mean Rank using the generated answers. We are the first to extend this paradigm to full two-way visual dialogue (2VD), where our model is capable of generating both questions and answers in sequence based on a visual input, for which we propose a set of novel evaluation measures and metrics.

This work addresses the task of camera localization in a known 3D scene given a single input RGB image. State-of-the-art approaches accomplish this in two steps: firstly, regressing for every pixel in the image its 3D scene coordinate and subsequently, using these coordinates to estimate the final 6D camera pose via RANSAC. To solve the first step. Random Forests (RFs) are typically used. On the other hand. Neural Networks (NNs) reign in many dense regression tasks, but are not test-time efficient. We ask the question: which of the two is best for camera localization? To address this, we make two method contributions: (1) a test-time efficient NN architecture which we term a ForestNet that is derived and initialized from a RF, and (2) a new fully-differentiable robust averaging technique for regression ensembles which can be trained end-to-end with a NN. Our experimental findings show that for scene coordinate regression, traditional NN architectures are superior to test-time efficient RFs and ForestNets, however, this does not translate to final 6D camera pose accuracy where RFs and ForestNets perform slightly better. To summarize, our best method, a ForestNet with a robust average, which has an equivalent fast and lightweight RF, improves over the state-of-the-art for camera localization on the 7-Scenes dataset [1]. While this work focuses on scene coordinate regression for camera localization, our innovations may also be applied to other continuous regression tasks.