This paper focuses on semi-supervised crowd counting, where only a small portion of the training data are labeled. We formulate the pixel-wise density value to regress as a probability distribution, instead of a single deterministic value. On this basis, we propose a semi-supervised crowd-counting model. Firstly, we design a pixel-wise distribution matching loss to measure the differences in the pixel-wise density distributions between the prediction and the ground truth; Secondly, we enhance the transformer decoder by using density tokens to specialize the forwards of decoders w.r.t. different density intervals; Thirdly, we design the interleaving consistency self-supervised learning mechanism to learn from unlabeled data efficiently. Extensive experiments on four datasets are performed to show that our method clearly outperforms the competitors by a large margin under various labeled ratio settings. Code will be released at https://github.com/LoraLinH/Semi-supervised-Counting-via-Pixel-by-pixel-Density-Distribution-Modelling.
Transformer has been popular in recent crowd counting work since it breaks the limited receptive field of traditional CNNs. However, since crowd images always contain a large number of similar patches, the self-attention mechanism in Transformer tends to find a homogenized solution where the attention maps of almost all patches are identical. In this paper, we address this problem by proposing Gramformer: a graph-modulated transformer to enhance the network by adjusting the attention and input node features respectively on the basis of two different types of graphs. Firstly, an attention graph is proposed to diverse attention maps to attend to complementary information. The graph is building upon the dissimilarities between patches, modulating the attention in an anti-similarity fashion. Secondly, a feature-based centrality encoding is proposed to discover the centrality positions or importance of nodes. We encode them with a proposed centrality indices scheme to modulate the node features and similarity relationships. Extensive experiments on four challenging crowd counting datasets have validated the competitiveness of the proposed method. Code is available at {https://github.com/LoraLinH/Gramformer}.
Coronary angiography remains the gold standard for diagnosis of coronary artery disease, the most common cause of death worldwide. While this procedure is performed more than 2 million times annually, there remain few methods for fast and accurate automated measurement of disease and localization of coronary anatomy. Here, we present our solution to the Automatic Region-based Coronary Artery Disease diagnostics using X-ray angiography images (ARCADE) challenge held at MICCAI 2023. For the artery segmentation task, our three-stage approach combines preprocessing and feature selection by classical computer vision to enhance vessel contrast, followed by an ensemble model based on YOLOv8 to propose possible vessel candidates by generating a vessel map. A final segmentation is based on a logic-based approach to reconstruct the coronary tree in a graph-based sorting method. Our entry to the ARCADE challenge placed 3rd overall. Using the official metric for evaluation, we achieved an F1 score of 0.422 and 0.4289 on the validation and hold-out sets respectively.
Coronary angiography continues to serve as the primary method for diagnosing coronary artery disease (CAD), which is the leading global cause of mortality. The severity of CAD is quantified by the location, degree of narrowing (stenosis), and number of arteries involved. In current practice, this quantification is performed manually using visual inspection and thus suffers from poor inter- and intra-rater reliability. The MICCAI grand challenge: Automatic Region-based Coronary Artery Disease diagnostics using the X-ray angiography imagEs (ARCADE) curated a dataset with stenosis annotations, with the goal of creating an automated stenosis detection algorithm. Using a combination of machine learning and other computer vision techniques, we propose the architecture and algorithm StenUNet to accurately detect stenosis from X-ray Coronary Angiography. Our submission to the ARCADE challenge placed 3rd among all teams. We achieved an F1 score of 0.5348 on the test set, 0.0005 lower than the 2nd place.
Federated Learning aims to learn a global model on the server side that generalizes to all clients in a privacy-preserving manner, by leveraging the local models from different clients. Existing solutions focus on either regularizing the objective functions among clients or improving the aggregation mechanism for the improved model generalization capability. However, their performance is typically limited by the dataset biases, such as the heterogeneous data distributions and the missing classes. To address this issue, this paper presents a cross-silo prototypical calibration method (FedCSPC), which takes additional prototype information from the clients to learn a unified feature space on the server side. Specifically, FedCSPC first employs the Data Prototypical Modeling (DPM) module to learn data patterns via clustering to aid calibration. Subsequently, the cross-silo prototypical calibration (CSPC) module develops an augmented contrastive learning method to improve the robustness of the calibration, which can effectively project cross-source features into a consistent space while maintaining clear decision boundaries. Moreover, the CSPC module's ease of implementation and plug-and-play characteristics make it even more remarkable. Experiments were conducted on four datasets in terms of performance comparison, ablation study, in-depth analysis and case study, and the results verified that FedCSPC is capable of learning the consistent features across different data sources of the same class under the guidance of calibrated model, which leads to better performance than the state-of-the-art methods. The source codes have been released at https://github.com/qizhuang-qz/FedCSPC.
Recently, the efficient deployment and acceleration of powerful vision transformers (ViTs) on resource-limited edge devices for providing multimedia services have become attractive tasks. Although early exiting is a feasible solution for accelerating inference, most works focus on convolutional neural networks (CNNs) and transformer models in natural language processing (NLP).Moreover, the direct application of early exiting methods to ViTs may result in substantial performance degradation. To tackle this challenge, we systematically investigate the efficacy of early exiting in ViTs and point out that the insufficient feature representations in shallow internal classifiers and the limited ability to capture target semantic information in deep internal classifiers restrict the performance of these methods. We then propose an early exiting framework for general ViTs termed LGViT, which incorporates heterogeneous exiting heads, namely, local perception head and global aggregation head, to achieve an efficiency-accuracy trade-off. In particular, we develop a novel two-stage training scheme, including end-to-end training and self-distillation with the backbone frozen to generate early exiting ViTs, which facilitates the fusion of global and local information extracted by the two types of heads. We conduct extensive experiments using three popular ViT backbones on three vision datasets. Results demonstrate that our LGViT can achieve competitive performance with approximately 1.8 $\times$ speed-up.
Timing synchronization (TS) is one of the key tasks in orthogonal frequency division multiplexing (OFDM) systems. However, multi-path uncertainty corrupts the TS correctness, making OFDM systems suffer from a severe inter-symbol-interference (ISI). To tackle this issue, we propose a timing-metric learning-based TS method assisted by a lightweight one-dimensional convolutional neural network (1-D CNN). Specifically, the receptive field of 1-D CNN is specifically designed to extract the metric features from the classic synchronizer. Then, to combat the multi-path uncertainty, we employ the varying delays and gains of multi-path (the characteristics of multi-path uncertainty) to design the timing-metric objective, and thus form the training labels. This is typically different from the existing timing-metric objectives with respect to the timing synchronization point. Our method substantively increases the completeness of training data against the multi-path uncertainty due to the complete preservation of metric information. By this mean, the TS correctness is improved against the multi-path uncertainty. Numerical results demonstrate the effectiveness and generalization of the proposed TS method against the multi-path uncertainty.
The binding problem is one of the fundamental challenges that prevent the artificial neural network (ANNs) from a compositional understanding of the world like human perception, because disentangled and distributed representations of generative factors can interfere and lead to ambiguity when complex data with multiple objects are presented. In this paper, we propose a brain-inspired hybrid neural network (HNN) that introduces temporal binding theory originated from neuroscience into ANNs by integrating spike timing dynamics (via spiking neural networks, SNNs) with reconstructive attention (by ANNs). Spike timing provides an additional dimension for grouping, while reconstructive feedback coordinates the spikes into temporal coherent states. Through iterative interaction of ANN and SNN, the model continuously binds multiple objects at alternative synchronous firing times in the SNN coding space. The effectiveness of the model is evaluated on synthetic datasets of binary images. By visualization and analysis, we demonstrate that the binding is explainable, soft, flexible, and hierarchical. Notably, the model is trained on single object datasets without explicit supervision on grouping, but successfully binds multiple objects on test datasets, showing its compositional generalization capability. Further results show its binding ability in dynamic situations.
In this paper, we propose a new agency-guided semi-supervised counting approach. First, we build a learnable auxiliary structure, namely the density agency to bring the recognized foreground regional features close to corresponding density sub-classes (agents) and push away background ones. Second, we propose a density-guided contrastive learning loss to consolidate the backbone feature extractor. Third, we build a regression head by using a transformer structure to refine the foreground features further. Finally, an efficient noise depression loss is provided to minimize the negative influence of annotation noises. Extensive experiments on four challenging crowd counting datasets demonstrate that our method achieves superior performance to the state-of-the-art semi-supervised counting methods by a large margin. Code is available.
In recent years, pre-trained models have become dominant in most natural language processing (NLP) tasks. However, in the area of Automated Essay Scoring (AES), pre-trained models such as BERT have not been properly used to outperform other deep learning models such as LSTM. In this paper, we introduce a novel multi-scale essay representation for BERT that can be jointly learned. We also employ multiple losses and transfer learning from out-of-domain essays to further improve the performance. Experiment results show that our approach derives much benefit from joint learning of multi-scale essay representation and obtains almost the state-of-the-art result among all deep learning models in the ASAP task. Our multi-scale essay representation also generalizes well to CommonLit Readability Prize data set, which suggests that the novel text representation proposed in this paper may be a new and effective choice for long-text tasks.