Video Description Model Based on Temporal-Spatial and Channel Multi-Attention Mechanisms

Video description plays an important role in the field of intelligent imaging technology. Attention perception mechanisms are extensively applied in video description models based on deep learning. Most existing models use a temporal-spatial attention mechanism to enhance the accuracy of models. Temporal attention mechanisms can obtain the global features of a video, whereas spatial attention mechanisms obtain local features. Nevertheless, because each channel of the convolutional neural network (CNN) feature maps has certain spatial semantic information, it is insufficient to merely divide the CNN features into regions and then apply a spatial attention mechanism. In this paper, we propose a temporal-spatial and channel attention mechanism that enables the model to take advantage of various video features and ensures the consistency of visual features between sentence descriptions to enhance the effect of the model. Meanwhile, in order to prove the effectiveness of the attention mechanism, this paper proposes a video visualization model based on the video description. Experimental results show that, our model has achieved good performance on the Microsoft Video Description (MSVD) dataset and a certain improvement on the Microsoft Research-Video to Text (MSR-VTT) dataset.

CREATING ENGINEERING TOPOGRAPHIC PLANS OF UNDERGROUND FACILITIES WITH THE VISUALIZATION ELEMENTS

The article considers the technological scheme of engineering geodetic works in designing and building underground utilities. The authors of this article proposed to conduct photo and video re-cording together with the surveying in order to increase informativity of engineering topographic plans, and describe relevancy and necessity of these works. They considered the experience of prac-tical use of hyperlinks in the software AutoCADCivil 3D, with the help of which it is possible to create photo and video visualization of the particular areas of underground utilities. This increases informativity of digital topographic plans and can simplify the work both on the stage of designing and on the stage of technical solutions in case of defects appearance in underground utilities.

Sinus Penetration of a Pulsating Device Versus the Classic Squeeze Bottle in Cadavers Undergoing Sinus Surgery

Objectives: Nasal irrigation is standard in the management of chronic rhinosinusitis both before and after surgical intervention. Numerous irrigation devices are commercially available. The aim of this study was to compare the efficacy of a handheld pulse irrigation device against the gold standard manual squeeze bottle after endoscopic sinus surgery (ESS). Methods: Five cadaveric specimens were prepared with video visualization ports into each sinus. Endoscopic sinus surgery was performed on each cadaver from minimal to maximal dissection. Sinuses were irrigated with fluorescein solution using both devices following each dissection. The irrigations were video recorded. A blinded independent observer scored each irrigation according to a defined scale. Results: Comparison of the 2 devices using an intraclass correlation coefficient (ICC = 0.39) showed the 2 systems differed. Observation of individual sinuses showed the squeeze bottle consistently provided greater irrigation in the maxillary ( P < .006), frontal ( P < .0001), and sphenoid ( P < .0001) sinuses. Pulse irrigation improved only in the maxillary sinus following ESS. Conclusion: The squeeze bottle consistently demonstrated superior irrigation in both native and operated conditions. Interestingly, saline penetration was not significantly improved after opening of the frontal and sphenoid sinuses. This is likely due to decreased backpressure in the nasal passage after opening the sinus cavities.