Software

UnifiedAR is a modular platform designed to recognize activities based on sensor data. With the proliferation of sensors in various devices and environments, there is a growing need to extract meaningful insights from the data they generate. Existing activity recognition systems often lack flexibility and interoperability, making it challenging to deploy across different domains and sensor types.

The selection of an optimal image segmentation technique relies heavily on the evaluation function utilized. However, recent investigations have revealed shortcomings in the performance evaluation of image segmentation models, particularly in the context of medical image segmentation where inter-voxel dependencies exist. Existing evaluation methods often suffer from a lack of robustness and exhibit statistical biases. Unlike conventional systems that classify predictions as either correct or incorrect, medical image segmentation predictions can exhibit both partial correctness and partial incorrectness simultaneously. This software addresses this inherent expressiveness and proposes a novel multi-modal evaluation (MME) method that aims to assess the effectiveness of various segmentation techniques in a comprehensive manner. The proposed MME method introduces a formal definition that incorporates interpretable criteria, including boundary alignment, detection, total volume, relative volume, and uniformity. By considering these salient properties of segmentation systems, the MME method provides a refined approach to segmentation assessment. By leveraging the MME method, researchers and practitioners can gain deeper insights into the efficiency of diverse segmentation techniques, particularly in the context of medical image analysis. The incorporation of interpretable criteria enhances the interpretability and applicability of segmentation evaluation, fostering advancements in the field of medical imaging and improving the overall quality of segmentation results.

The "OpenCV Jupyter UI" project addresses the compatibility issue between OpenCV's user interface components and Jupyter Notebooks. In remote Jupyter environments such as Jupyter Notebook or Google Colab, the traditional method of using `cv2.imshow` for displaying images is not supported. This project introduces an alternative solution by providing the `jcv2.imshow` function, which is compatible with Jupyter environments. By replacing `cv2.imshow` with `jcv2.imshow`, users can seamlessly display images within Jupyter Notebooks without encountering compatibility errors. The `jcv2.imshow` function leverages Jupyter's capabilities to render images and ensures a smooth integration between OpenCV and Jupyter. Additionally, the project addresses the need for user interaction and replaces the usage of `cv2.waitKey` with `jcv2.waitKey`. For instance, `jcv2.waitKey(1000)` waits for a button press for one second. By bridging the gap between OpenCV and Jupyter, the "OpenCV Jupyter UI" project enhances the usability and convenience of using OpenCV's user interface components within Jupyter Notebooks. Users can seamlessly leverage OpenCV functionalities for image processing and visualization, empowering them to work efficiently in Jupyter environments.

ConfJournalRank isintegrates conference and journal information from multiple authoritative sources. The system collects data from various reputable platforms, including Scopus, Scimago, Web of Science, WikiCFP, Core Ranking, Qualis, ERA, research.com, and Open Research, and consolidates this information into a unified platform. By gathering data from diverse sources, ConfJournalRank provides users with a centralized location to access and explore conference and journal details. Users no longer need to visit multiple websites or platforms to obtain relevant information. Instead, they can conveniently access a wide range of conference and journal data through a single interface. The system offers a comprehensive view of conferences and journals, incorporating various metrics, rankings, and categorizations from each source. This enables users to compare and assess the significance, impact, and quality of conferences and journals based on multiple criteria. ConfJournalRank aims to streamline the process of gathering conference and journal information, making it more efficient and accessible for researchers, scholars, and other users in the academic community. By providing a consolidated platform, the system facilitates informed decision-making and enhances the overall research experience for users seeking up-to-date and reliable conference and journal information.

This library presents a real-time timer designed for profiling Python functions or code snippets within the Jupyter environment. The timer integrates seamlessly with Jupyter widgets, providing an interactive and extendable tree-based visualization of profiling results. Key features of the proposed timer include the ability to filter out external library profiling, allowing users to focus solely on their own code. Additionally, the timer incorporates a threshold-based filter to exclude functions with very short execution times, enabling users to concentrate on more significant performance concerns. Furthermore, the timer supports variable depth analysis, facilitating the identification of time-consuming functions within nested call hierarchies. It also handles loop or multiple function calls, providing comprehensive profiling capabilities for iterative or repetitive code structures. Moreover, recursive function calls are appropriately handled, ensuring accurate and insightful profiling results. To optimize efficiency, users have the option to globally disable the timer by setting Profiler.GlobalDisable to True, thereby saving valuable execution time when profiling is not required. Overall, this real-time timer with interactive Jupyter widgets and an extendable tree-based interface empowers users to effectively profile Python functions or code snippets, filter results based on specific criteria, and gain deeper insights into their code's performance characteristics.

Système pour apprendre un ensemble de clauses du premier ordre à partir d'exemples ambigus

Extension du logiciel Galicia pour la construction incrémentale et la fusion de treillis Alpha

Enregistrement numéro IDDN.FR.001.490019.000S.P.2008.000.20000 auprès de l'agence de protection logiciel