Existing panoramic layout estimation solutions tend to recover room boundaries from a vertically compressed sequence, yielding imprecise results as the compression process often muddles the semantics between various planes. Besides, these datadriven approaches impose an urgent demand for massive data annotations, which are laborious and time-consuming. For the first problem, we propose an orthogonal plane disentanglement network (termed DOPNet) to distinguish ambiguous semantics. DOPNet consists of three modules that are integrated to deliver distortion-free, semantics-clean, and detail-sharp disentangled representations, which benefit the subsequent layout recovery. For the second problem, we present an unsupervised adaptation technique tailored for horizon-depth and ratio representations. Concretely, we introduce an optimization strategy for decision-level layout analysis and a 1D cost volume construction method for feature-level multi-view aggregation, both of which are designed to fully exploit the geometric consistency across multiple perspectives. The optimizer provides a reliable set of pseudo-labels for network training, while the 1D cost volume enriches each view with comprehensive scene information derived from other perspectives. Extensive experiments demonstrate that our solution outperforms other SoTA models on both monocular layout estimation and multi-view layout estimation tasks.

Depressive disorders are common among people with epilepsy (PwE). We here aimed to report an unbiased automatic classification of epilepsy comorbid depressive disorder cases via training a linear support vector machine (SVM) model using the interictal heart rate variability (HRV) data. One hundred and eighty-six subjects participated in this study. Among all participants, we recorded demographic information, epilepsy states and neuropsychiatric features. For each subject, we performed simultaneous electrocardiography and electroencephalography recordings both in wakefulness and non-rapid eye movement (NREM) sleep stage. Using these data, we systematically explored the full parameter space in order to determine the most effective combinations of data to classify the depression status in PwE. PwE with depressive disorders exhibited significant alterations in HRV parameters, including decreased time domain and nonlinear domain values both in wakefulness and NREM sleep stage compared with without depressive disorders and non-epilepsy controls. Interestingly, PwE without depressive disorder showed the same level of HRV values as the non-epilepsy control subjects. The SVM classification model of PwE depression status achieved a higher classification accuracy with the combination of HRV parameters in wakefulness and NREM sleep stage. Furthermore, the receiver operating characteristic (ROC) curve of the SVM classification model showed a satisfying area under the ROC curve (AUC: 0.758). Intriguingly, we found that the HRV measurements during NREM sleep are particularly important for correct classification, suggesting a mechanistic link between the dysregulation of heart rate during sleep and the development of depressive disorders in PwE. Our classification model may provide an objective measurement to assess the depressive status in PwE.

No abstract

No abstract