An Empirical Study of GPT-3 for Few-Shot Knowledge-Based VQA

Knowledge-based visual question answering (VQA) involves answering questions that require external knowledge not present in the image. Existing methods first retrieve knowledge from external resources, then reason over the selected knowledge, the input image, and question for answer prediction. However, this two-step approach could lead to mismatches that potentially limit the VQA performance. For example, the retrieved knowledge might be noisy and irrelevant to the question, and the re-embedded knowledge features during reasoning might deviate from their original meanings in the knowledge base (KB). To address this challenge, we propose PICa, a simple yet effective method that Prompts GPT3 via the use of Image Captions, for knowledge-based VQA. Inspired by GPT-3’s power in knowledge retrieval and question answering, instead of using structured KBs as in previous work, we treat GPT-3 as an implicit and unstructured KB that can jointly acquire and process relevant knowledge. Specifically, we first convert the image into captions (or tags) that GPT-3 can understand, then adapt GPT-3 to solve the VQA task in a few-shot manner by just providing a few in-context VQA examples. We further boost performance by carefully investigating: (i) what text formats best describe the image content, and (ii) how in-context examples can be better selected and used. PICa unlocks the first use of GPT-3 for multimodal tasks. By using only 16 examples, PICa surpasses the supervised state of the art by an absolute +8.6 points on the OK-VQA dataset. We also benchmark PICa on VQAv2, where PICa also shows a decent few-shot performance.

LAViTeR: Learning Aligned Visual and Textual Representations Assisted by Image and Caption Generation

Pre-training visual and textual representations from large-scale image-text pairs is becoming a standard approach for many downstream vision-language tasks. The transformer-based models learn inter and intra-modal attention through a list of self-supervised learning tasks. This paper proposes LAViTeR, a novel architecture for visual and textual representation learning. The main module, Visual Textual Alignment (VTA) will be assisted by two auxiliary tasks, GAN-based image synthesis and Image Captioning. We also propose a new evaluation metric measuring the similarity between the learnt visual and textual embedding. The experimental results on two public datasets, CUB and MS-COCO, demonstrate superior visual and textual representation alignment in the joint feature embedding space

Detection of GAN-synthesized street videos

Research on the detection of AI-generated videos has focused almost exclusively on face videos, usually referred to as deepfakes. Manipulations like face swapping, face reenactment and expression manipulation have been the subject of an intense research with the development of a number of efficient tools to distinguish artificial videos from genuine ones. Much less attention has been paid to the detection of artificial non-facial videos. Yet, new tools for the generation of such kind of videos are being developed at a fast pace and will soon reach the quality level of deepfake videos. The goal of this paper is to investigate the detectability of a new kind of AI-generated videos framing driving street sequences (here referred to as DeepStreets videos), which, by their nature, can not be analysed with the same tools used for facial deepfakes. Specifically, we present a simple frame-based detector, achieving very good performance on state-of-the-art DeepStreets videos generated by the Vid2vid architecture. Noticeably, the detector retains very good performance on compressed videos, even when the compression level used during training does not match that used for the test videos.

Unsupervised Change Detection in Hyperspectral Images using Feature Fusion Deep Convolutional Autoencoders

Binary change detection in bi-temporal co-registered hyperspectral images is a challenging task due to a large number of spectral bands present in the data. Researchers, therefore, try to handle it by reducing dimensions. The proposed work aims to build a novel feature extraction system using a feature fusion deep convolutional autoencoder for detecting changes between a pair of such bi-temporal co-registered hyperspectral images. The feature fusion considers features across successive levels and multiple receptive fields and therefore adds a competitive edge over the existing feature extraction methods. The change detection technique described is completely unsupervised and is much more elegant than other supervised or semi-supervised methods which require some amount of label information. Different methods have been applied to the extracted features to find the changes in the two images and it is found that the proposed method clearly outperformed the state of the art methods in unsupervised change detection for all the datasets.

Panoptic Narrative Grounding

This paper proposes Panoptic Narrative Grounding, a spatially fine and general formulation of the natural language visual grounding problem. We establish an experimental framework for the study of this new task, including new ground truth and metrics, and we propose a strong baseline method to serve as stepping stone for future work. We exploit the intrinsic semantic richness in an image by including panoptic categories, and we approach visual grounding at a fine-grained level by using segmentations. In terms of ground truth, we propose an algorithm to automatically transfer Localized Narratives annotations to specific regions in the panoptic segmentations of the MS COCO dataset. To guarantee the quality of our annotations, we take advantage of the semantic structure contained in WordNet to exclusively incorporate noun phrases that are grounded to a meaningfully related panoptic segmentation region. The proposed baseline achieves a performance of 55.4 absolute Average Recall points. This result is a suitable foundation to push the envelope further in the development of methods for Panoptic Narrative Grounding.

LibFewShot: A Comprehensive Library for Few-shot Learning

Few-shot learning, especially few-shot image classification, has received increasing attention and witnessed significant advances in recent years. Some recent studies implicitly show that many generic techniques or “tricks”, such as data augmentation, pre-training, knowledge distillation, and self-supervision, may greatly boost the performance of a few-shot learning method. Moreover, different works may employ different software platforms, different training schedules, different backbone architectures and even different input image sizes, making fair comparisons difficult and practitioners struggle with reproducibility. To address these situations, we propose a comprehensive library for few-shot learning (LibFewShot) by re-implementing seventeen state-of-the-art few-shot learning methods in a unified framework with the same single codebase in PyTorch. Furthermore, based on LibFewShot, we provide comprehensive evaluations on multiple benchmark datasets with multiple backbone architectures to evaluate common pitfalls and effects of different training tricks. In addition, given the recent doubts on the necessity of meta- or episodic-training mechanism, our evaluation results show that such kind of mechanism is still necessary especially when combined with pre-training. We hope our work can not only lower the barriers for beginners to work on few-shot learning but also remove the effects of the nontrivial tricks to facilitate intrinsic research on few-shot learning. The source code is available from

Negative Sample Matters: A Renaissance of Metric Learning for Temporal Grounding

Temporal grounding aims to temporally localize a video moment in the video whose semantics are related to a given natural language query. Existing methods typically apply a detection or regression pipeline on the fused representation with a focus on designing complicated heads and fusion strategies. Instead, from a perspective on temporal grounding as a metric-learning problem, we present a Dual Matching Network (DMN), to directly model the relations between language queries and video moments in a joint embedding space. This new metric-learning framework enables fully exploiting negative samples from two new aspects: constructing negative cross-modal pairs from a dual matching scheme and mining negative pairs across different videos. These new negative samples could enhance the joint representation learning of two modalities via cross-modal pair discrimination to maximize their mutual information. Experiments show that DMN achieves highly competitive performance compared with state-of-the-art methods on four video grounding benchmarks. Based on DMN, we present a winner solution for STVG challenge of the 3rd PIC workshop. This suggests that metric-learning is still a promising method for temporal grounding via capturing the essential cross-modal correlation in a joint embedding space.

Spatio-Temporal Recurrent Networks for Event-Based Optical Flow Estimation

Event camera has offered promising alternative for visual perception, especially in high speed and high dynamic range scenes. Recently, many deep learning methods have shown great success in providing model-free solutions to many event-based problems, such as optical flow estimation. However, existing deep learning methods did not address the importance of temporal information well from the perspective of architecture design and cannot effectively extract spatio-temporal features. Another line of research that utilizes Spiking Neural Network suffers from training issues for deeper architecture. To address these points, a novel input representation is proposed that captures the events temporal distribution for signal enhancement. Moreover, we introduce a spatio-temporal recurrent encoding-decoding neural network architecture for event-based optical flow estimation, which utilizes Convolutional Gated Recurrent Units to extract feature maps from a series of event images. Besides, our architecture allows some traditional frame-based core modules, such as correlation layer and iterative residual refine scheme, to be incorporated. The network is end-to-end trained with self-supervised learning on the Multi-Vehicle Stereo Event Camera dataset. We have shown that it outperforms all the existing state-of-the-art methods by a large margin.

Temporally Coherent Person Matting Trained on Fake-Motion Dataset

We propose a novel neural-network-based method to perform matting of videos depicting people that does not require additional user input such as trimaps. Our architecture achieves temporal stability of the resulting alpha mattes by using motion-estimation-based smoothing of image-segmentation algorithm outputs, combined with convolutional-LSTM modules on U-Net skip connections.

TADA: Taxonomy Adaptive Domain Adaptation

Traditional domain adaptation addresses the task of adapting a model to a novel target domain under limited or no additional supervision. While tackling the input domain gap, the standard domain adaptation settings assume no domain change in the output space. In semantic prediction tasks, different datasets are often labeled according to different semantic taxonomies. In many real-world settings, the target domain task requires a different taxonomy than the one imposed by the source domain. We therefore introduce the more general taxonomy adaptive domain adaptation (TADA) problem, allowing for inconsistent taxonomies between the two domains. We further propose an approach that jointly addresses the image-level and label-level domain adaptation. On the label-level, we employ a bilateral mixed sampling strategy to augment the target domain, and a relabelling method to unify and align the label spaces. We address the image-level domain gap by proposing an uncertainty-rectified contrastive learning method, leading to more domain-invariant and class discriminative features. We extensively evaluate the effectiveness of our framework under different TADA settings: open taxonomy, coarse-to-fine taxonomy, and partially-overlapping taxonomy. Our framework outperforms previous state-of-the-art by a large margin, while capable of adapting to new target domain taxonomies.