arXiv subjects

Diogo Mendonça

Publications and source records attributed to Diogo Mendonça.

4 recordsLinked to original sources

arXiv2026

Seg2Track++: Probabilistic Track Validation and Data Association for Multi-Object Tracking and Segmentation

Autonomous systems require robust Multi-Object Tracking and Segmentation (MOTS) to operate reliably in dynamic environments, ensuring consistent object identities and precise mask-level delineation. Foundation models such as SAM2 have shown strong zero-shot generalization for segmentation, but their direct application to MOTS is limited by unreliable track association and false-positive propagation. This work introduces Seg2Track++, a framework that integrates instance segmentation with SAM2 and a novel track management module to perform zero-shot MOTS with enhanced temporal consistency. Tracks are associated using Mask Centroid Distance (MCD) and Confidence-Aware Cost Modulation (CCM), while Probabilistic Track Validation (PTV) employs a Bernoulli filter to validate track existence and suppress ghost tracks. Experimental results on KITTI MOTS demonstrate improved identity preservation, reduced false-positive propagation, and robust track management without fine-tuning.

cs.CV↗

arXiv2026

Seg2Track-SAM2: SAM2-based Multi-object Tracking and Segmentation

Autonomous-driving perception systems require robust Multi-Object Tracking (MOT) to operate reliably in dynamic environments. MOT maintains consistent object identities across frames while preserving spatial accuracy. Recent foundation models, such as SAM2, provide promptable video segmentation without task-specific fine-tuning. However, their direct application to Multi-Object Tracking and Segmentation (MOTS) remains limited by the absence of explicit identity management mechanisms and by growing memory requirements during tracking. This work introduces Seg2Track-SAM2, a framework that integrates pretrained object detectors with SAM2 and a dedicated Seg2Track module to support track initialization, data association, and track refinement. The method operates without dataset-specific fine-tuning and remains detector-agnostic. Experimental evaluation on the KITTI MOTS and MOTS Challenge benchmarks shows that Seg2Track-SAM2 ranks fourth overall in both datasets while achieving the highest association accuracy (AssA) among compared methods. In addition, a sliding-window memory strategy reduces memory usage by up to 75% with minimal impact on tracking performance, enabling deployment under resource constraints. Together, these results indicate that Seg2Track-SAM2 improves identity consistency and memory efficiency in MOTS without requiring dataset-specific training. The code is available at https://github.com/hcmr-lab/Seg2Track-SAM2.

cs.CV↗

arXiv2022

Assessment of nanoparticle immersion depth at liquid interfaces from chemically equivalent macroscopic surfaces

Hypothesis: We test whether the wettability of nanoparticles (NPs) straddling at an air/water surface or oil/water interface can be extrapolated from sessile drop-derived macroscopic contact angles (mCAs) on planar substrates, assuming that both the nanoparticles and the macroscopic substrates are chemically equivalent and feature the same electrokinetic potential. Experiments: Pure silica (SiO2) and amino-terminated silica (APTES-SiO2) NPs are compared to macroscopic surfaces with extremely low roughness (root mean square [RMS] roughness <= 2 nm) or a roughness determined by a close-packed layer of NPs (RMS roughness about 35 nm). Equivalence of the surface chemistry is assessed by comparing the electrokinetic potentials of the NPs via electrophoretic light scattering and of the macroscopic substrates via streaming current analysis. The wettability of the macroscopic substrates is obtained from advancing (ACAs) and receding contact angles (RCAs) and in situ synchrotron X-ray reflectivity (XRR) provided by the NP wettability at the liquid interfaces. Findings: Generally, the RCA on smooth surfaces provides a good estimate of NP wetting properties. However, mCAs alone cannot predict adsorption barriers that prevent NP segregation to the interface, as is the case with the pure SiO2 nanoparticles. This strategy greatly facilitates assessing the wetting properties of NPs for applications such as emulsion formulation, flotation, or water remediation.

physics.chem-ph↗

arXiv2021

Cataloging Dependency Injection Anti-Patterns in Software Systems

Context: Dependency Injection (DI) is a commonly applied mechanism to decouple classes from their dependencies in order to provide higher modularization. However, bad DI practices often lead to negative consequences, such as increasing coupling. Although white literature conjectures about the existence of DI anti-patterns, there is no evidence on their practical relevance, usefulness, and generality. Objective: The objective of this study is to propose and evaluate a catalog of Java DI anti-patterns and associated refactorings. Methodology: We reviewed existing reported DI anti-patterns in order to analyze their completeness. The limitations found in literature motivated proposing a novel catalog of 12 DI anti-patterns. We developed a tool to statically analyze the occurrence level of the candidate DI anti-patterns in both open-source and industry projects. Next, we survey practitioners to assess their perception on the relevance, usefulness, and their willingness on refactoring anti-pattern instances of the catalog. Results: Our static code analyzer tool showed a relative recall of 92.19% and high average precision. It revealed that at least 9 different DI anti-patterns appeared frequently in the analyzed projects. Besides, our survey confirmed the perceived relevance of the catalog and developers expressed their willingness to refactor instances of anti-patterns from source code. Conclusion: The catalog contains Java DI anti-patterns that occur in practice and that are perceived as useful. Sharing it with practitioners may help them to avoid such anti-patterns, thus improving source-code quality.

cs.SE↗