This paper presents a comprehensive image labelling workflow that integrates the labelme software with Grounded-SAM2. Our approach aims to enhance the efficiency and precision of semantic- and instance segmentation annotation processes. First, labelme is used to create new annotations, forming a foundational dataset for fine-tuning a SAM2.1 model. Then, after training has been completed, a more specialised model is ready to be used for annotation. This iterative process enables us to refine the model's performance and generate high-quality annotations for a diverse range of images. Our approach is specifically designed to operate locally on decent GPUs, ensuring users can efficiently execute the whole workflow inside a single application without relying on cloud-based resources. This enhances accessibility and reduces latency in the annotation process. Furthermore, our workflow keeps annotation data under the user's control, ensuring data privacy and enabling users to securely manage their sensitive information throughout the labelling process.

The creation of digital twins of cityscapes requires the understanding the semantics of relevant objects encountered in the scene, with classes possibly not well covered in standard datasets. In this paper, we assess the generalization capabilities of models for semantic segmentation of street scenes trained on synthetic data to real images. In order to perform assessment on real data, we create a dataset from images obtained from a vehicle-based capture setup, and propose a highly automated annotation pipeline for generating ground truth. We train three state of the art segmentation models on our own synthetic dataset, and evaluate the methods on real data.

Music is a multifaceted art form that requires a nuanced and comprehensive framework for analysis. This framework should encompass correlations among diverse musical attributes, including melody, harmony, rhythm, timbre, and dynamics. However, modeling the interaction between these attributes is anticipated to require substantial computational resources. In this study, we propose a Bayesian model to represent various musical attributes and explore its implementation and inference on a quantum computer. We show how to translate the Bayesian model to a quantum circuit and perform inference using the quantum rejection sampling algorithm on a quantum simulator. Through empirical evaluation utilizing a real-world dataset, we compare the performance of the quantum algorithm against classical rejection sampling. Our results reveal that the quantum method improves the acceptance rate of generated samples, thereby enhancing posterior probability estimates.

Augmented Reality (AR) applications rely heavily on the accurate detection and segmentation of objects in order to seamlessly integrate virtual content into real world environments. However, achieving robust object detection and segmentation in AR scenes remains challenging, in particular when specific classes need to be covered. Due to the lack of sufficient real annotated data, training can rely on synthetic data, which has the issue of the domain gap between synthetic and real-world images. This paper addresses the challenge of generalisation in object detection/instance segmentation models trained with synthetic data. It presents techniques to improve the generalisation capability of object detection models trained on synthetic data for real-world applications. Our method is based on the model soup approach, where models trained with different data subsets and hyperparameters are combined to achieve better generalisation performance. The experimental results on the validation dataset of the ADAPT sim2real Object Detection Challenge 2023 demonstrate the effectiveness of our approach in achieving good generalisation performance. This marks a significant step towards more reliable and adaptable object detection and segmentation approaches for AR applications.

We present a C++ library implementing part 2, 3, 4 and 5 of the MPEG 7 multimedia content description standard, including the updated version finalized in 2004. The library supports handling of MPEG-7 descriptions as trees of typed objects, supporting (de)serialization from/to XML. It has convenient and powerful features such as creation of subtrees by XPath statements and is extensible at runtime. The library is available for Windows, Linux/Unix and Mac OS X. It has been provided under a free use license for several years, downloaded more than 5,200 times and used in a large number of projects. It has been published under GNU LGPL in 2009. This paper discusses the key functionalities of the library as well as some exemplary applications.

A considerable amount of work in larger film and video restoration projects is dedicated to manually exploring the audiovisual content in order to estimate the costs for restoration and to plan the restoration. Manual exploration is a significant cost factor. In this paper we propose automatic content analysis algorithms and summarization techniques which allow the reduction of manual inspection time in a software based restoration environment. The throughput requirement for analysis of dust and other defects is reached by sparse application of the detectors in the image sequence while retaining sufficient detection accuracy. Analysis result metadata are represented in a MPEG-7 standard compliant way. The proposed defect summary visualization tools facilitate efficient exploration of visually impaired content by the user.