Flow Estimation Research Articles

EventHDR: From Event to High-Speed HDR Videos and Beyond.

Event cameras are innovative neuromorphic sensors that asynchronously capture the scene dynamics. Due to the event-triggering mechanism, such cameras record event streams with much shorter response latency and higher intensity sensitivity compared to conventional cameras. On the basis of these features, previous works have attempted to reconstruct high dynamic range (HDR) videos from events, but have either suffered from unrealistic artifacts or failed to provide sufficiently high frame rates. In this paper, we present a recurrent convolutional neural network that reconstruct high-speed HDR videos from event sequences, with a key frame guidance to prevent potential error accumulation caused by the sparse event data. Additionally, to address the problem of severely limited real dataset, we develop a new optical system to collect a real-world dataset with paired high-speed HDR videos and event streams, facilitating future research in this field. Our dataset provides the first real paired dataset for event-to-HDR reconstruction, avoiding potential inaccuracies from simulation strategies. Experimental results demonstrate that our method can generate high-quality, high-speed HDR videos. We further explore the potential of our work in cross-camera reconstruction and downstream computer vision tasks, including object detection, panoramic segmentation, optical flow estimation, and monocular depth estimation under HDR scenarios.

Leveraging Crowdsourced Activity Information for Transit Stations Flow Estimation

Leveraging Crowdsourced Activity Information for Transit Stations Flow Estimation

Self-Supervised Cyclic Diffeomorphic Mapping for Soft Tissue Deformation Recovery in Robotic Surgery Scenes.

The ability to recover tissue deformation from visual features is fundamental for many robotic surgery applications. This has been a long-standing research topic in computer vision, however, is still unsolved due to complex dynamics of soft tissues when being manipulated by surgical instruments. The ambiguous pixel correspondence caused by homogeneous texture makes achieving dense and accurate tissue tracking even more challenging. In this paper, we propose a novel self-supervised framework to recover tissue deformations from stereo surgical videos. Our approach integrates semantics, cross-frame motion flow, and long-range temporal dependencies to enable the recovered deformations to represent actual tissue dynamics. Moreover, we incorporate diffeomorphic mapping to regularize the warping field to be physically realistic. To comprehensively evaluate our method, we collected stereo surgical video clips containing three types of tissue manipulation (i.e., pushing, dissection and retraction) from two different types of surgeries (i.e., hemicolectomy and mesorectal excision). Our method has achieved impressive results in capturing deformation in 3D mesh, and generalized well across manipulations and surgeries. It also outperforms current state-of-the-art methods on non-rigid registration and optical flow estimation. To the best of our knowledge, this is the first work on self-supervised learning for dense tissue deformation modeling from stereo surgical videos. Our code will be released.

Low-light optical flow estimation with hidden feature supervision using a Siamese network

Low-light optical flow estimation with hidden feature supervision using a Siamese network

Adaptive Boundary Observers for Hyperbolic PDEs With Application to Traffic Flow Estimation

Adaptive Boundary Observers for Hyperbolic PDEs With Application to Traffic Flow Estimation

Active Scene Flow Estimation for Autonomous Driving via Real-Time Scene Prediction and Optimal Decision

Active Scene Flow Estimation for Autonomous Driving via Real-Time Scene Prediction and Optimal Decision

Dynamic-guided Spatiotemporal Attention for Echocardiography Video Segmentation.

Left ventricle (LV) endocardium segmentation in echocardiography video has received much attention as an important step in quantifying LV ejection fraction. Most existing methods are dedicated to exploiting temporal information on top of 2D convolutional networks. In addition to single appearance semantic learning, some research attempted to introduce motion cues through the optical flow estimation (OFE) task to enhance temporal consistency modeling. However, OFE in these methods is tightly coupled to LV endocardium segmentation, resulting in noisy inter-frame flow prediction, and post-optimization based on these flows accumulates errors. To address these drawbacks, we propose dynamic-guided spatiotemporal attention (DSA) for semi-supervised echocardiography video segmentation. We first fine-tune the off-the-shelf OFE network RAFT on echocardiography data to provide dynamic information. Taking inter-frame flows as additional input, we use a dual-encoder structure to extract motion and appearance features separately. Based on the connection between dynamic continuity and semantic consistency, we propose a bilateral feature calibration module to enhance both features. For temporal consistency modeling, the DSA is proposed to aggregate neighboring frame context using deformable attention that is realized by offsets grid attention. Dynamic information is introduced into DSA through a bilateral offset estimation module to effectively combine with appearance semantics and predict attention offsets, thereby guiding semantic-based spatiotemporal attention. We evaluated our method on two popular echocardiography datasets, CAMUS and EchoNet-Dynamic, and achieved state-of-the-art.

Event-Based Optical Flow via Transforming Into Motion-Dependent View.

Event cameras respond to temporal dynamics, helping to resolve ambiguities in spatio-temporal changes for optical flow estimation. However, the unique spatio-temporal event distribution challenges the feature extraction, and the direct construction of motion representation through the orthogonal view is less than ideal due to the entanglement of appearance and motion. This paper proposes to transform the orthogonal view into a motion-dependent one for enhancing event-based motion representation and presents a Motion View-based Network (MV-Net) for practical optical flow estimation. Specifically, this motion-dependent view transformation is achieved through the Event View Transformation Module, which captures the relationship between the steepest temporal changes and motion direction, incorporating these temporal cues into the view transformation process for feature gathering. This module includes two phases: extracting the temporal evolution clues by central difference operation in the extraction phase and capturing the motion pattern by evolution-guided deformable convolution in the perception phase. Besides, the MV-Net constructs an eccentric downsampling process to avoid response weakening from the sparsity of events in the downsampling stage. The whole network is trained end-to-end in a self-supervised manner, and the evaluations conducted on four challenging datasets reveal the superior performance of the proposed model compared to state-of-the-art (SOTA) methods.

An Artificial Intelligence Prediction Approach for Behavioral Intentions of Health Tourism: a Protection Motivation Theory-based Perspective

Abstract This study applies an artificial intelligence (AI) method, informed by the Protection Motivation Theory (PMT), to predict the behavioral intentions of tourists in a healthy town in Yunnan. This study looks at online search data to guess when a lot of tourists will come by combining text mining with the SPCA-LSTM model. This model combines seasonal and trend decomposition using Loess (STL) with Long Short-Term Memory (LSTM) networks. The model is more accurate than traditional forecasting methods and provides a daily average tourist flow estimate of 3,247 with minimal prediction errors. The average absolute error of 806.4074 and the root mean square error (RMSE) of 959.775 further highlight the model’s performance. This research contributes significantly to tourism management and strategic planning, particularly in health-related destinations. The model provides a reliable benchmark for predicting tourist flows enhancing decision-making processes in the tourism sector.

A Deep Learning Model for Precipitation Nowcasting Using Multiple Optical Flow Algorithms

Abstract The optical flow technique has advantages in motion tracking and has long been employed in precipitation nowcasting to track the motion of precipitation fields using ground radar datasets. However, the performance and forecast time scale of models based on optical flow are limited. Here, we present the results of the application of the deep learning method to optical flow estimation to extend its forecast time scale and enhance the performance of nowcasting. It is shown that a deep learning model can better capture both multispatial and multitemporal motions of precipitation events compared with traditional optical flow estimation methods. The model comprises two components: 1) a regression process based on multiple optical flow algorithms, which more accurately captures multispatial features compared with a single optical flow algorithm; and 2) a U-Net-based network that trains multitemporal features of precipitation movement. We evaluated the model performance with cases of precipitation in South Korea. In particular, the regression process minimizes errors by combining multiple optical flow algorithms with a gradient descent method and outperforms other models using only a single optical flow algorithm up to a 3-h lead time. Additionally, the U-Net plays a crucial role in capturing nonlinear motion that cannot be captured by a simple advection model through traditional optical flow estimation. Consequently, we suggest that the proposed optical flow estimation method with deep learning could play a significant role in improving the performance of current operational nowcasting models, which are based on traditional optical flow methods. Significance Statement The purpose of this study is to improve the accuracy of short-term rainfall prediction based on optical flow methods that have been employed for operational precipitation nowcasting. By utilizing open-source libraries, such as OpenCV, and commonly applied machine learning techniques, such as multiple linear regression and U-Net networks, we propose an accessible model for enhancing prediction accuracy. We expect that the improvement in prediction accuracy will significantly improve the practical application of operational precipitation nowcasting.

FedRSU: Federated Learning for Scene Flow Estimation on Roadside Units

FedRSU: Federated Learning for Scene Flow Estimation on Roadside Units

A Deep Graph Neural Network Approach for Assessing Origin Destination Traffic Flow Estimates Based on COVID-19 Data

A Deep Graph Neural Network Approach for Assessing Origin Destination Traffic Flow Estimates Based on COVID-19 Data

Hydrological models and datasets to support flood flow estimation at river-bridge intersections

A working group on the "Hydraulic Compatibility of Bridges" (sites.google.com/view/gii-ponti) has been recently established in the hydraulic engineering academic community aiming at reviewing methodologies, studying good practices, and suggesting guidelines for assessing the bridge hydraulic compatibility. In this initiative, a subgroup is focusing on the hydrological aspects as in Italy there are not standard procedures to estimate design flood values requested by the Italian technical standards of constructions (NTC 2018) over the whole country as, for example, in the UK where the Flood Estimation Handbook provides standard data and methods. Some models are available at the regional scale but are often based on not up-to-date data; moreover, while in large rivers flood discharge values, defined according to the European Flood Directive have been computed and mapped, in small river catchments they are not available.In this context, the unit of Politecnico di Torino working on hydrology is going to release some country-wide hydrological datasets, including all the available official information on discharge and rainfall extremes updated to recent years, that can be used as a common reference for bridge analyses. These include high-resolution maps of extreme precipitation statistics, a database of flood flow extremes recorded at more than 600 river sections, and a catalogue of river catchments characteristics (morphology, climate, land use, etc).Nation-wide datasets can support the application of standardized hydrological analyses, needed for flood hazard assessment, especially in small basins. Operational products will be made available within the PNRR RETURN project in order to be easily available and referable for any hydrological analysis in Italy.

Dense Optical Flow Estimation Using Sparse Regularizers From Reduced Measurements

Dense Optical Flow Estimation Using Sparse Regularizers From Reduced Measurements

Point Cloud Densification Based on Scene Flow Estimation and Kalman Refinement

Point Cloud Densification Based on Scene Flow Estimation and Kalman Refinement

Multi-Person Pose Tracking With Sparse Key-Point Flow Estimation and Hierarchical Graph Distance Minimization.

In this paper, we propose a novel framework for multi-person pose estimation and tracking on challenging scenarios. In view of occlusions and motion blurs which hinder the performance of pose tracking, we proposed to model humans as graphs and perform pose estimation and tracking by concentrating on the visible parts of human bodies which are informative about complete skeletons under incomplete observations. Specifically, the proposed framework involves three parts: (i) A Sparse Key-point Flow Estimating Module (SKFEM) and a Hierarchical Graph Distance Minimizing Module (HGMM) for estimating pixel-level and human-level motion, respectively; (ii) Pixel-level appearance consistency and human-level structural consistency are combined in measuring the visibility scores of body joints. The scores guide the pose estimator to predict complete skeletons by observing high-visibility parts, under the assumption that visible and invisible parts are inherently correlated in human part graphs. The pose estimator is iteratively fine-tuned to achieve this capability; (iii) Multiple historical frames are combined to benefit tracking which is implemented using HGMM. The proposed approach not only achieves state-of-the-art performance on PoseTrack datasets but also contributes to significant improvements in other tasks such as human-related anomaly detection.

Extreme Flow Estimation in Ungauged Basins: Case of Eastern Black Sea

Taşkınlar, dünya genelinde can ve mal kayıpları açısından incelendiğinde depremlerin ardından en yüksek kayba neden olan doğal afetlerdir. Taşkınları önlemek ve taşkınlardan korunmak için yapılan çalışmalarda kullanılan en temel veri akarsu debileridir. Akarsu debilerinden istatistiksel yöntemler aracılığıyla taşkın debileri hesaplanabilmektedir. Dünyanın farklı yerlerinde olduğu gibi ülkemizde de akım verisi açısından fakir sahalar bulunmaktadır. Bu sahalarda yer alan havzalara ait akımların tahmini farklı yöntemlerle yapılabilmektedir. Bu çalışmada bahsedilen yöntemlerden birisi olan L-momentler yöntemi ile Doğu Karadeniz Havzasında yer alan alt havzaların taşkın debilerinin hesaplanması amaçlanmıştır. Bu amaç doğrultusunda havzadaki akım gözlem istasyonlarına bölgesel analizler uygulanmış, en uygun bölgesel dağılım belirlenmiş ve bu dağılıma uygun yöntemlerle farklı tekrar periyotlarına sahip taşkın debileri hesaplanmıştır. Çalışmanın çıktıları hem hidrolojik hem de hidrodinamik çalışmalar için altlık oluşturmaktadır.

Regional Flood Frequency Analysis in the Doce River Basin and Coastal Regions in Espírito Santo State, Brazil – Evaluation of the Performance of L-Moments Approach

Objective: To evaluate the performance of the Index-Flood methods using L-Moments when analyzing the regional flood frequency. Theoretical benchmark: The maximum stream flows estimation is relevant for projects and reduction of flood impacts. However, hydrologic regionalization methods are usually employed due to the absence of maximum stream flow data. In this context, it is relevant to understand the methods performance of regional analysis available. Method: Regional flood frequency analisys in the Doce River basin and in basins located predominantly in the state of Espírito Santo by the Index-Flood methods, using L-Moments, direct multiple regression and Dimensionless Curve. Results and conclusion: The Index-Flood method using L-Moments presented better results than the other methods of regionalization evaluated. Changes in stream flow series and configuration of homogeneous regions did not alter significantly the results obtained with the Index-Flood method using L-Moments. Implications of research: Improvement of the process of regional analysis of flood frequency. Originality/value: Consistent analysis of the frequency of occurrence of floods, a recurrent socio-environmental problem aggravated in the context of climate change, constitutes a relevant advance for the management of water resources and has the potential to assist in the establishment of public policies aimed at territorial planning.

Dynamic probabilistic analytical modeling for estimating rainfall–runoff transformation rates in drylands

Abstract Precipitation-runoff changes used in the development of precipitation-runoff models depict general runoff mechanisms in physical precipitation-runoff processes. For this purpose, it must be able to adequately express the characteristics of the physical system. Similar to other water resources management models, analytical-probabilistic models may be developed with different levels of complexity according to different types of rainfall–runoff developments. In this research, a log-normal probabilistic model was used to estimate the return period of rainfall and Soil and Water Assessment Tool (SWAT) software was used to convert rainfall values into discharge and runoff height in agricultural lands. The slope and the level of plant shade cover were considered the two basic factors in the estimation of the subsurface flow used in agriculture. Slopes of less than 0.015 and land cover level of more than 70% of extreme points were obtained to ensure subsurface water sources and prevent erosion.

Effect of Teleconnection Patterns on the Formation of Potential Ecological Flow Variables in Lowland Rivers

Climate is probably the most important factor affecting river discharge and flow dynamics. Low flows in rivers during the warm period cause stress to aquatic ecosystems and pose a challenge to sustainable water management. Previous research has shown that the average minimum discharge of the 30 driest continuous days, known as Q30, is a suitable measure for ecological flow estimation in Lithuania. This study aims to examine whether large-scale atmospheric processes, so-called teleconnections, can have an impact on Q30 during the warm period. Hydrological data for 1961–2020 from 25 water gauging stations were used to search for hydrological response signals with five selected climate indices (NAO, SCA, POL, EA/WR, and EA). Pearson correlation and Wilcoxon–Mann–Whitney test approaches were applied. The results suggested that the EA/WR and NAO had the strongest influence on Q30 in the studied region during the warm period. The positive phases of the indices tended to cause a greater decrease in Q30 values due to the prevailing easterly edge of the anticyclonic circulation over the studied region determined by the EA/WR and NAO indices, while the negative phases of the mentioned indices caused an increase and greater dispersion of Q30.