Demian Inostroza, Ekaterina Vylomova, Charles Kemp, Mae Carroll, Wanchun Li, Meladel Mistica. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Large-size very-high-resolution (VHR) remote sensing imagery has emerged as a critical data source for high-precision vector mapping of multi-scale geographical elements such as building, water, road and etc. When dealing with the large-size image, due to the limited memory of GPU, the deep learning-based vector mapping methods often employ the sliding block strategy. This inevitably leads to the degenerated performance because of the stitching difficulty of the sliding blocks' vector mapping results. Therefore, it is necessary to conduct full-scope vector mapping via mining the consistent cue in large-size remote sensing imagery. To this end, this paper presents a novel global context-aware local point optimization method. To leverage the global context, this paper proposes a novel pyramid fusion network (PFNet) to conduct semantic segmentation of the large-size image in an end-to-end manner. Under the constraint of the global semantic segmentation result, a new inflection-point perception network (IPNet) is proposed to generate a set of stable points to depict the boundary of each element. Extensive experiments on building, water and road datasets, where each image has over 100 million pixels, show that our method obviously outperforms the existing methods.

With the increasing resolution of remote sensing imagery (RSI), large-size RSI has emerged as a vital data source for high-precision vector mapping of geographic objects. Existing methods are typically constrained to processing small image patches, which often leads to the loss of contextual information and produces fragmented vector outputs. To address these, this paper introduces HoliTracer, the first framework designed to holistically extract vectorized geographic objects from large-size RSI. In HoliTracer, we enhance segmentation of large-size RSI using the Context Attention Net (CAN), which employs a local-to-global attention mechanism to capture contextual dependencies. Furthermore, we achieve holistic vectorization through a robust pipeline that leverages the Mask Contour Reformer (MCR) to reconstruct polygons and the Polygon Sequence Tracer (PST) to trace vertices. Extensive experiments on large-size RSI datasets, including buildings, water bodies, and roads, demonstrate that HoliTracer outperforms state-of-the-art methods. Our code and data are available in https://github.com/vvangfaye/HoliTracer.

Global surface water detection in very-high-resolution (VHR) satellite imagery can directly serve major applications such as refined flood mapping and water resource assessment. Although achievements have been made in detecting surface water in small-size satellite images corresponding to local geographic scales, datasets and methods suitable for mapping and analyzing global surface water have yet to be explored. To encourage the development of this task and facilitate the implementation of relevant applications, we propose the GLH-water dataset that consists of 250 satellite images and 40.96 billion pixels labeled surface water annotations that are distributed globally and contain water bodies exhibiting a wide variety of types (e.g. , rivers, lakes, and ponds in forests, irrigated fields, bare areas, and urban areas). Each image is of the size 12,800 × 12,800 pixels at 0.3 meter spatial resolution. To build a benchmark for GLH-water, we perform extensive experiments employing representative surface water detection models, popular semantic segmentation models, and ultra-high resolution segmentation models. Furthermore, we also design a strong baseline with the novel pyramid consistency loss (PCL) to initially explore this challenge, increasing IoU by 2.4% over the next best baseline. Finally, we implement the cross-dataset generalization and pilot area application experiments, and the superior performance illustrates the strong generalization and practical application value of GLH-water dataset. Project page: https://jack-bo1220.github.io/project/GLH-water.html

A key feature of extended target (ET) is that it can produce multiple measurements, thus providing detailed information such as size, orientation. However, due to the measurement number decrease, the ET tends to be a point target (PT) when it becomes far from the sensor, leading to disability of estimating the extensions. In this paper we consider the problem of distributed ETs tracking based on multiple sensors. In such a case the measurements decrease of an ET can be compensated by other sensors, so that the tracking performance can be maintained. In the proposed algorithm the targets are modeled by Poisson multi-Bernoulli (PMB) random finite set (RFS) and the state of each target consists of two parts representing the possibility of being ET and PT, respectively. In the local filtering state, interaction between ET and PT states is considered in the prediction step for target identity change between ET and PT, based on which the local PMB filter is achieved for seamlessly tracking ETs and PTs. In the fusion part a generalized covariance intersection (GCI) based criterion is proposed to fuse the posteriors of each sensor. The performance of proposed algorithm is verified via simulations.

Target state estimation with Doppler-only sensors has attracted a lot of attention due to its wide potential applications in target localization and tracking. While existing Doppler-only tracking methods rely on the assumption that Doppler sensors have been correctly registered, in many practical cases there can be significant registration errors which imply measurement biases and thus performance degradation in target state estimation. Motivated by this issue, the present paper addresses the problem of jointly estimating target state and sensor biases based on Doppler-only measurements. The proposed method consists of two phases, i.e., (1) raw estimation of the target state without considering sensor biases, followed by (2) a bias compensation step that relies on linearization of the measurement function and joint estimation of target state-sensor biases via a least square method. The Cramer-Rao lower bound (CRLB) in estimating sensor biases is evaluated and the performance of the proposed method is also assessed via simulations.

In this study, for the problem of estimating the location and speed of a moving target in the non‐coherent multiple‐input multiple‐output (MIMO) radar systems with widely separated antennas, the authors propose two new methods, in which the parameters used are the joint of bearing, elevation, frequency‐of‐arrival (FOA) and time‐of‐arrival (TOA). The two proposed methods are based on non‐coherent MIMO radar systems, but method 1 centralises all measuring parameters in one linear equation and processes together, while method 2 divides the measurements into several groups according to the different transmitter–receiver pairs. In this study, the authors assume that bearing, elevation, FOA and TOA parameters have already been measured by a preprocessing algorithm. For the both methods, an initial guess is acquired through a best linear unbiased estimator. Then for method 1, a more explicit solution can be acquired by employing a maximum likelihood estimator for decorrelation, while the method 2 applies maximum likelihood estimation of a first‐order Taylor expansion for a better solution. The simulations show that these two methods are effective and both of them can attain to Cramér–Rao lower bound at sufficiently moderate noise conditions.

Web applications can suffer from poor reliability, and AJAX technology makes Web sites even more error-prone. Failures of a Web application, particularly user-visible failures, impact users' satisfaction and may drive users away from using the Web site. Conventional testing techniques are inadequate for improving AJAX applications' reliability, and application providers commonly rely on fast failure detection, which is challenging. In this paper, we present a novel technique for automatically detecting user-visible failures in AJAX applications. Our technique trains a Bayesian model to analyze users' interaction behaviors to infer whether such user responses are related to user-visible failures. We implemented our technique in a tool called SIRANA. We performed a case study using a commercial AJAX application with seeded bugs, and collected users' interaction data during 14 one-hour sessions. We evaluated our technique using SIRANA applied to the collected data. The results demonstrate the effectiveness of our technique: It not only detected all seeded bugs, but also detected four real, previously-unknown bugs.

This paper presents a method that addresses two practical issues concerning the use of random test selection for regression testing: the number of random samples needed from the test suite to provide reliable results, and the confidence levels of the predictions made by the random samples. The method applies the Chernoff bound, which has been applied in various randomized algorithms, to compute the error bound for random test selection. The paper presents three example applications, based on the method, for regression testing. The main benefits of the method are that it requires no distribution information about the test suite from which the samples are taken, and the computation of the confidence level is independent of the size of the test suite. The paper also presents the results of an empirical evaluation of the technique on a set of C programs, which have been used in many testing experiments, along with three of the GCC compilers. The results demonstrate the effectiveness of the method and show its potential for regression testing on real-world, large-scale applications.

Dynamic taint analysis is gaining momentum. Techniques based on dynamic tainting have been successfully used in the context of application security, and now their use is also being explored in different areas, such as program understanding, software testing, and debugging. Unfortunately, most existing approaches for dynamic tainting are defined in an ad-hoc manner, which makes it difficult to extend them, experiment with them, and adapt them to new contexts. Moreover, most existing approaches are focused on data-flow based tainting only and do not consider tainting due to control flow, which limits their applicability outside the security domain. To address these limitations and foster experimentation with dynamic tainting techniques, we defined and developed a general framework for dynamic tainting that (1) is highly flexible and customizable, (2) allows for performing both data-flow and control-flow based tainting conservatively, and (3) does not rely on any customized run-time system. We also present DYTAN, an implementation of our framework that works on x86 executables, and a set of preliminary studies that show how DYTAN can be used to implement different tainting-based approaches with limited effort. In the studies, we also show that DYTAN can be used on real software, by using FIREFOX as one of our subjects, and illustrate how the specific characteristics of the tainting approach used can affect efficiency and accuracy of the taint analysis, which further justifies the use of our framework to experiment with different variants of an approach.

Visualising large graphs faces the challenges of both data complexity and visual complexity. This paper presents a framework for visualising large graphs that reduces data complexity using the clustered graph model and provides users with navigational approaches for browsing clustered graphs. A key design task of such a system is to define a strategy for generating logical abstractions of a clustered graph during navigation. An appropriate abstraction strategy should represent a clustered graph well and avoid visual overload. The semantic fisheye view of a clustered graph is proposed for such a purpose. Two case studies were investigated, and the experiment results show that during navigation the first-order fisheye view of a clustered graph conserves visual complexity at a constant level.

Visualizing software architecture faces the challenges of both data complexity and visual complexity. This paper presents an approach for visualizing software architecture, which reduces data complexity using the clustered graph model and navigates pictures of clustered graphs with constant visual complexity. A graph drawing algorithm is introduced to generate visualizations of clustered graphs. A semantic fisheye view of a clustered graph is proposed for conserving constant visual complexity. Animation is used to present smooth transition of visualizations. A case study is investigated to navigate the architecture of the Compiler c488.