Author: Fabrizio Falchi;

Text-to-Motion Retrieval: Towards Joint Understanding of Human Motion Data and Natural Language

Due to recent advances in pose-estimation methods, human motion can be extracted from a common video in the form of 3D skeleton sequences. Despite wonderful application opportunities, effective and efficient content-based access to large volumes of such spatio-temporal skeleton data still remains a challenging problem. In this paper, we propose a novel content-based text-to-motion retrieval task, which aims at retrieving relevant motions based on a specified natural-language textual description. To define baselines for this uncharted task, we employ the BERT and CLIP language representations to encode the text modality and successful spatio-temporal models to encode the motion modality. We additionally introduce our transformer-based approach, called Motion Transformer (MoT), which employs divided space-time attention to effectively aggregate the different skeleton joints in space and time. Inspired by the recent progress in text-to-image/video matching, we experiment with two widely-adopted metric-learning loss functions. Finally, we set up a common evaluation protocol by defining qualitative metrics for assessing the quality of the retrieved motions, targeting the two recently-introduced KIT Motion-Language and HumanML3D datasets. The code for reproducing our results is available here: https://github.com/mesnico/text-to-motion-retrieval.

Escaping local minima in deep reinforcement learning for video summarization

State-of-the-art deep neural unsupervised video summarization methods mostly fall under the adversarial reconstruction framework. This employs a Generative Adversarial Network (GAN) structure and Long Short-Term Memory (LSTM) auto-encoders during its training stage. The typical result is a selector LSTM that sequentially receives video frame representations and outputs corresponding scalar importance factors, which are then used to select key-frames. This basic approach has been augmented with an additional Deep Reinforcement Learning (DRL) agent, trained using the Discriminator’s output as a reward, which learns to optimize the selector’s outputs. However, local minima are a well-known problem in DRL. Thus, this paper presents a novel regularizer for escaping local loss minima, in order to improve unsupervised key-frame extraction. It is an additive loss term employed during a second training phase, that rewards the difference of the neural agent’s parameters from those of a previously found good solution. Thus, it encourages the training process to explore more aggressively the parameter space in order to discover a better local loss minimum. Evaluation performed on two public datasets shows considerable increases over the baseline and against the state-of-the-art.

Cross-Forgery Analysis of Vision Transformers and CNNs for Deepfake Image Detection

Deepfake Generation Techniques are evolving at a rapid pace, making it possible to create realistic manipulated images and videos and endangering the serenity of modern society. The continual emergence of new and varied techniques brings with it a further problem to be faced, namely the ability of deepfake detection models to update themselves promptly in order to be able to identify manipulations carried out using even the most recent methods. This is an extremely complex problem to solve, as training a model requires large amounts of data, which are difficult to obtain if the deepfake generation method is too recent. Moreover, continuously retraining a network would be unfeasible. In this paper, we ask ourselves if, among the various deep learning techniques, there is one that is able to generalise the concept of deepfake to such an extent that it does not remain tied to one or more specific deepfake generation methods used in the training set. We compared a Vision Transformer with an EfficientNetV2 on a cross-forgery context based on the ForgeryNet dataset. From our experiments, It emerges that EfficientNetV2 has a greater tendency to specialize often obtaining better results on training methods while Vision Transformers exhibit a superior generalization ability that makes them more competent even on images generated with new methodologies.

An Optimized Pipeline for Image-Based Localization in Museums from Egocentric Images

With the increasing interest in augmented and virtual reality, visual localization is acquiring a key role in many downstream applications requiring a real-time estimate of the user location only from visual streams. In this paper, we propose an optimized hierarchical localization pipeline by specifically tackling cultural heritage sites with specific applications in museums. Specifically, we propose to enhance the Structure from Motion (SfM) pipeline for constructing the sparse 3D point cloud by a-priori filtering blurred and near-duplicated images. We also study an improved inference pipeline that merges similarity-based localization with geometric pose estimation to effectively mitigate the effect of strong outliers. We show that the proposed optimized pipeline obtains the lowest localization error on the challenging Bellomo dataset [11]. Our proposed approach keeps both build and inference times bounded, in turn enabling the deployment of this pipeline in real-world scenarios.

MC-GTA: A Synthetic Benchmark for Multi-Camera Vehicle Tracking

Multi-camera vehicle tracking (MCVT) aims to trace multiple vehicles among videos gathered from overlapping and non-overlapping city cameras. It is beneficial for city-scale traffic analysis and management as well as for security. However, developing MCVT systems is tricky, and their real-world applicability is dampened by the lack of data for training and testing computer vision deep learning-based solutions. Indeed, creating new annotated datasets is cumbersome as it requires great human effort and often has to face privacy concerns. To alleviate this problem, we introduce MC-GTA – Multi Camera Grand Tracking Auto, a synthetic collection of images gathered from the virtual world provided by the highly-realistic Grand Theft Auto 5 (GTA) video game. Our dataset has been recorded from several cameras recording urban scenes at various crossroads. The annotations, consisting of bounding boxes localizing the vehicles with associated unique IDs consistent across the video sources, have been automatically generated by interacting with the game engine. To assess this simulated scenario, we conduct a performance evaluation using an MCVT SOTA approach, showing that it can be a valuable benchmark that mitigates the need for real-world data. The MC-GTA dataset and the code for creating new ad-hoc custom scenarios are available at https://github.com/GaetanoV10/GT5-Vehicle-BB.

ALADIN: Distilling Fine-grained Alignment Scores for Efficient Image-Text Matching and Retrieval

Image-text matching is gaining a leading role among tasks involving the joint understanding of vision and language. In literature, this task is often used as a pre-training objective to forge architectures able to jointly deal with images and texts. Nonetheless, it has a direct downstream application: cross-modal retrieval, which consists in finding images related to a given query text or vice-versa. Solving this task is of critical importance in cross-modal search engines. Many recent methods proposed effective solutions to the image-text matching problem, mostly using recent large vision-language (VL) Transformer networks. However, these models are often computationally expensive, especially at inference time. This prevents their adoption in large-scale cross-modal retrieval scenarios, where results should be provided to the user almost instantaneously. In this paper, we propose to fill in the gap between effectiveness and efficiency by proposing an ALign And DIstill Network (ALADIN). ALADIN first produces high-effective scores by aligning at fine-grained level images and texts. Then, it learns a shared embedding space – where an efficient kNN search can be performed – by distilling the relevance scores obtained from the fine-grained alignments. We obtained remarkable results on MS-COCO, showing that our method can compete with state-of-the-art VL Transformers while being almost 90 times faster. The code for reproducing our results is available at https://github.com/mesnico/ALADIN.

Recurrent Vision Transformer for Solving Visual Reasoning Problems

Although convolutional neural networks (CNNs) showed remarkable results in many vision tasks, they are still strained by simple yet challenging visual reasoning problems. Inspired by the recent success of the Transformer network in computer vision, in this paper, we introduce the Recurrent Vision Transformer (RViT) model. Thanks to the impact of recurrent connections and spatial attention in reasoning tasks, this network achieves competitive results on the same-different visual reasoning problems from the SVRT dataset. The weight-sharing both in spatial and depth dimensions regularizes the model, allowing it to learn using far fewer free parameters, using only 28k training samples. A comprehensive ablation study confirms the importance of a hybrid CNN + Transformer architecture and the role of the feedback connections, which iteratively refine the internal representation until a stable prediction is obtained. In the end, this study can lay the basis for a deeper understanding of the role of attention and recurrent connections for solving visual abstract reasoning tasks. The code for reproducing our results is publicly available here: https://tinyurl.com/recvit.

 

Deep Features for CBIR with Scarce Data using Hebbian Learning

Features extracted from Deep Neural Networks (DNNs) have proven to be very effective in the context of Content Based Image Retrieval (CBIR). Recently, biologically inspired Hebbian learning algorithms have shown promises for DNN training. In this contribution, we study the performance of such algorithms in the development of feature extractors for CBIR tasks. Specifically, we consider a semi-supervised learning strategy in two steps: first, an unsupervised pre-training stage is performed using Hebbian learning on the image dataset; second, the network is fine-tuned using supervised Stochastic Gradient Descent (SGD) training. For the unsupervised pre-training stage, we explore the nonlinear Hebbian Principal Component Analysis (HPCA) learning rule. For the supervised fine-tuning stage, we assume sample efficiency scenarios, in which the amount of labeled samples is just a small fraction of the whole dataset. Our experimental analysis, conducted on the CIFAR10 and CIFAR100 datasets, shows that, when few labeled samples are available, our Hebbian approach provides relevant improvements compared to various alternative methods.