Contrastively-trained Vision-Language Models (VLMs), such as CLIP, have become the standard approach for learning discriminative vision-language representations. However, these models often exhibit shallow language understanding, manifesting bag-of-words behaviour. These limitations are reinforced by their dual-encoder design, which induces a modality gap. Additionally, the reliance on vast web-collected data corpora for training makes the process computationally expensive and introduces significant privacy concerns. To address these limitations, in this work, we challenge the necessity of vision encoders for retrieval tasks by introducing a \textit{vision-free, single-encoder} retrieval pipeline. Departing from the traditional text-to-image retrieval paradigm, we migrate to a text-to-text paradigm with the assistance of VLLM-generated structured image descriptions. We demonstrate that this paradigm shift has significant advantages, including a substantial reduction of the modality gap, improved compositionality, and better performance on short and long caption queries, all attainable with only two hours of calibration on two GPUs. Additionally, substituting raw images with textual descriptions introduces a more privacy-friendly alternative for retrieval. To further assess generalisation and address some of the shortcomings of prior compositionality benchmarks, we release two benchmarks derived from Flickr30k and COCO, containing diverse compositional queries made of short captions, which we coin subFlickr and subCOCO. Our vision-free retriever matches and often surpasses traditional multimodal models. Importantly, our approach achieves state-of-the-art zero-shot performance on multiple retrieval and compositionality benchmarks, with models as small as 0.3B parameters.

Recognising emotions in context involves identifying an individual’s apparent emotions while considering contextual cues from the surrounding scene. Previous approaches to this task have typically designed explicit scene-encoding architectures or incorporated external scene-related information, such as captions. However, these methods often utilise limited contextual information or rely on intricate training pipelines to decouple noise from relevant information. In this work, we leverage the capabilities of Vision-and-Large-Language Models (VLLMs) to enhance in-context emotion classification in a more straightforward manner. Our proposed method follows a simple yet effective two-stage approach. First, we prompt VLLMs to generate natural language descriptions of the subject’s apparent emotion in relation to the visual context. Second, the descriptions, along with the visual input, are used to train a transformer-based architecture that fuses text and visual features before the final classification task. This method not only simplifies the training process but also significantly improves performance. Experimental results demonstrate that the textual descriptions effectively guide the model to constrain the noisy visual input, allowing our fused architecture to outperform individual modalities. Our approach achieves state-of-the-art performance across three datasets, BoLD, EMOTIC, and CAER-S, without bells and whistles. Our code will be made publicly available.

In this paper, we observe that a straight bipartite matching loss can be applied to the output tokens of a vision transformer. This results in a backbone + MLP architecture that can do both tasks without the need of an extra encoder-decoder head and learnable queries. We show that a single MViTv2-S architecture trained with bipartite matching to perform both tasks surpasses the same MViTv2-S when trained with RoI align on pre-computed bounding boxes. With a careful design of token pooling and the proposed training pipeline, our Bipartite-Matching Vision Transformer model, BMViT, achieves +3 mAP on AVA2.2. w.r.t. the two-stage MViTv2-S counterpart. Code is available at https://github.com/IoannaNti/BMViT Action Localization is a challenging problem that combines detection and recognition tasks, which are often addressed separately. State-of-the-art methods rely on off-the-shelf bounding box detections pre-computed at high resolution, and propose transformer models that focus on the classification task alone. Such two-stage solutions are prohibitive for real-time deployment. On the other hand, single-stage methods target both tasks by devoting part of the network (generally the backbone) to sharing the majority of the work-load, compromising performance for speed. These methods build on adding a DETR head with learnable queries that after cross-and self-attention can be sent to corresponding MLPs for detecting a person's bounding box and action. However, DETR-like architectures are challenging to train and can incur in big complexity.

This paper is on long-term video understanding where the goal is to recognise human actions over long temporal windows (up to minutes long). In prior work, long temporal context is captured by constructing a long-term memory bank consisting of past and future video features which are then integrated into standard (short-term) video recognition backbones through the use of attention mechanisms. Two well-known problems related to this approach are the quadratic complexity of the attention operation and the fact that the whole feature bank must be stored in memory for inference. To address both issues, we propose an alternative to attention-based schemes which is based on a low-rank approximation of the memory obtained using Singular Value Decomposition. Our scheme has two advantages: (a) it reduces complexity by more than an order of magnitude, and (b) it is amenable to an efficient implementation for the calculation of the memory bases in an incremental fashion which does not require the storage of the whole feature bank in memory. The proposed scheme matches or surpasses the accuracy achieved by attention-based mechanisms while being memory-efficient. Through extensive experiments, we demonstrate that our framework generalises to different architectures and tasks, outperforming the state-of-the-art in three datasets.

Action Units (AUs) are geometrically-based atomic facial muscle movements known to produce appearance changes at specific facial locations. Motivated by this observation we propose a novel AU modelling problem that consists of jointly estimating their localisation and intensity. To this end, we propose a simple yet efficient approach based on Heatmap Regression that merges both problems into a single task. A Heatmap models whether an AU occurs or not at a given spatial location. To accommodate the joint modelling of AUs intensity, we propose variable size heatmaps, with their amplitude and size varying according to the labelled intensity. Using Heatmap Regression, we can inherit from the progress recently witnessed in facial landmark localisation. Building upon the similarities between both problems, we devise a transfer learning approach where we exploit the knowledge of a network trained on large-scale facial landmark datasets. In particular, we explore different alternatives for transfer learning through a) fine-tuning, b) adaptation layers, c) attention maps, and d) reparametrisation. Our approach effectively inherits the rich facial features produced by a strong face alignment network, with minimal extra computational cost. We empirically validate that our system sets a new state-of-the-art on three popular datasets, namely BP4D, DISFA, and FERA2017.