Abstract: Transformer models have shown great success handling long-range interactions, making them a promising tool for modeling video. However, they lack inductive biases and scale quadratically with input length. These limitations are further exacerbated when dealing with the high dimensionality introduced by the temporal dimension. While there are surveys analyzing the advances of Transformers for vision, none focus on an in-depth analysis of video-specific designs. In this survey, we analyze the main contributions and trends of works leveraging Transformers to model video. Specifically, we delve into how videos are handled at the input level first. Then, we study the architectural changes made to deal with video more efficiently, reduce redundancy, re-introduce useful inductive biases, and capture long-term temporal dynamics. In addition, we provide an overview of different training regimes and explore effective self-supervised learning strategies for video. Finally, we conduct a performance comparison on the most common benchmark for Video Transformers (i.e., action classification), finding them to outperform 3D ConvNets even with less computational complexity.

Abstract: Multimodal learning from document data has achieved great success lately as it allows to pre-train semantically meaningful features as a prior into a learnable downstream approach. In this paper, we approach the document classification problem by learning cross-modal representations through language and vision cues, considering intra- and inter-modality relationships. Instead of merging features from different modalities into a common representation space, the proposed method exploits high-level interactions and learns relevant semantic information from effective attention flows within and across modalities. The proposed learning objective is devised between intra- and inter-modality alignment tasks, where the similarity distribution per task is computed by contracting positive sample pairs while simultaneously contrasting negative ones in the common feature representation space}. Extensive experiments on public document classification datasets demonstrate the effectiveness and the generalization capacity of our model on both low-scale and large-scale datasets.

Abstract: Deep neural networks represent a compelling technique to tackle complex real-world problems, but are over-parameterized and often suffer from over- or under-confident estimates. Deep ensembles have shown better parameter estimations and often provide reliable uncertainty estimates that contribute to the robustness of the results. In this work, we propose a new metric to identify samples that are hard to classify. Our metric is defined as coincidence score for deep ensembles which measures the agreement of its individual models. The main hypothesis we rely on is that deep learning algorithms learn the low-loss samples better compared to large-loss samples. In order to compensate for this, we use controlled over-sampling on the identified ”hard” samples using proper data augmentation schemes to enable the models to learn those samples better. We validate the proposed metric using two public food datasets on different backbone architectures and show the improvements compared to the conventional deep neural network training using different performance metrics.

Abstract: Multi-label zero-shot learning strives to classify images into multiple unseen categories for which no data is available during training. The test samples can additionally contain seen categories in the generalized variant. Existing approaches rely on learning either shared or label-specific attention from the seen classes. Nevertheless, computing reliable attention maps for unseen classes during inference in a multi-label setting is still a challenge. In contrast, state-of-the-art single-label generative adversarial network (GAN) based approaches learn to directly synthesize the class-specific visual features from the corresponding class attribute embeddings. However, synthesizing multi-label features from GANs is still unexplored in the context of zero-shot setting. When multiple objects occur jointly in a single image, a critical question is how to effectively fuse multi-class information. In this work, we introduce different fusion approaches at the attribute-level, feature-level and cross-level (across attribute and feature-levels) for synthesizing multi-label features from their corresponding multi-label class embeddings. To the best of our knowledge, our work is the first to tackle the problem of multi-label feature synthesis in the (generalized) zero-shot setting. Our cross-level fusion-based generative approach outperforms the state-of-the-art on three zero-shot benchmarks: NUS-WIDE, Open Images and MS COCO. Furthermore, we show the generalization capabilities of our fusion approach in the zero-shot detection task on MS COCO, achieving favorable performance against existing methods.

Abstract: Vascular tortuosity of supra-aortic vessels is widely considered one of the main reasons for failure and delays in endovascular treatment of large vessel occlusion in patients with acute ischemic stroke. Characterization of tortuosity is a challenging task due to the lack of objective, robust and effective analysis tools. We present a fully automatic method for arterial segmentation, vessel labelling and tortuosity feature extraction applied to the supra-aortic region. A sample of 566 computed tomography angiography scans from acute ischemic stroke patients (aged 74.8 ± 12.9, 51.0% females) were used for training, validation and testing of a segmentation module based on a U-Net architecture (162 cases) and a vessel labelling module powered by a graph U-Net (566 cases). Successively, 30 cases were processed for testing of a tortuosity feature extraction module. Measurements obtained through automatic processing were compared to manual annotations from two observers for a thorough validation of the method. The proposed feature extraction method presented similar performance to the inter-rater variability observed in the measurement of 33 geometrical and morphological features of the arterial anatomy in the supra-aortic region. This system will contribute to the development of more complex models to advance the treatment of stroke by adding immediate automation, objectivity, repeatability and robustness to the vascular tortuosity characterization of patients.

Abstract: Despite its wide applications in criminal investigations and clinical communications with patients suffering from autism, automatic micro-expression recognition remains a challenging problem because of the lack of training data and imbalanced classes problems. In this study, we proposed a meta-learning-based multi-model fusion network (Meta-MMFNet) to solve the existing problems. The proposed method is based on the metric-based meta-learning pipeline, which is specifically designed for few-shot learning and is suitable for model-level fusion. The frame difference and optical flow features were fused, deep features were extracted from the fused feature, and finally in the meta-learning-based framework, weighted sum model fusion method was applied for micro-expression classification. Meta-MMFNet achieved better results than state-of-the-art methods on four datasets. The code is available at https://github.com/wenjgong/meta-fusion-based-method.

Abstract: Abstract
Purpose:
Carotid revascularization can be associated with modifications of the vascular geometry, which may lead to complications. The changes on the vessel angulation before and after a carotid WallStent (WS) implantation are compared against 2 new dual-layer devices, CGuard (CG) and RoadSaver (RS).
Materials and Methods:
The study prospectively recruited 217 consecutive patients (112 GC, 73 WS, and 32 RS, respectively). Angiography projections were explored and the one having a higher arterial angle was selected as a basal view. After stent implantation, a stent control angiography was performed selecting the projection having the maximal angle. The same procedure is followed in all the 3 stent types to guarantee comparable conditions. The angulation changes on the stented segments were quantified from both angiographies. The statistical analysis quantitatively compared the pre-and post-angles for the 3 stent types. The results are qualitatively illustrated using boxplots. Finally, the relation between pre- and post-angles measurements is analyzed using linear regression.
Results:
For CG, no statistical difference in the axial vessel geometry between the basal and postprocedural angles was found. For WS and RS, statistical difference was found between pre- and post-angles. The regression analysis shows that CG induces lower changes from the original curvature with respect to WS and RS.
Conclusion:
Based on our results, CG determines minor changes over the basal morphology than WS and RS stents. Hence, CG respects better the native vessel anatomy than the other stents.
Level of Evidence: Level 4, Case Series.

Abstract: Neural networks can be successfully used for cross-component prediction in video coding. In particular, attention-based architectures are suitable for chroma intra prediction using luma information because of their capability to model relations between difierent channels. However, the complexity of such methods is still very high and should be further reduced, especially for decoding. In this paper, a cost-effective attention-based neural network is designed for chroma intra prediction. Moreover, with the goal of further improving coding performance, a novel approach is introduced to utilize more boundary information effectively. In addition to improving prediction, a simplification methodology is also proposed to reduce inference complexity by simplifying convolutions. The proposed schemes are integrated into H.266/Versatile Video Coding (VVC) pipeline, and only one additional binary block-level syntax flag is introduced to indicate whether a given block makes use of the proposed method. Experimental results demonstrate that the proposed scheme achieves up to −0.46%/−2.29%/−2.17% BD-rate reduction on Y/Cb/Cr components, respectively, compared with H.266/VVC anchor. Reductions in the encoding and decoding complexity of up to 22% and 61%, respectively, are achieved by the proposed scheme with respect to the previous attention-based chroma intra prediction method while maintaining coding performance.