Find open source projects actively accepting contributors. Search repositories, filter by program milestones, difficulty tags, or tech stack.
Use our Orbit AI Matcher to find out! Get instant matching scores based on your developer skills, preferred frameworks, and contribution experience.
Convert your selected open-source project into a winning GSoC, LFX, or Outreachy application using Proposal Studio.
Apache NuttX already supports ELF loading and shared objects, but it still lacks a practical lifecycle layer for installing, updating, versioning, and rolling back applications after the base firmware is deployed. This project will build a thin distribution helper on top of existing NuttX services instead of introducing a full package-manager ecosystem. The work will target a supported ESP32-S3 board with PSRAM and use sim for reproducible early validation. The implementation will reuse existing NuttX subsystems such as NSH, filesystems, networking, and BINFMT/ELF loading, while adding the missing pieces: local metadata, versioned application storage, direct dependency handling, staged activation, rollback, and shared-library reuse rules. The main deliverables are: reproducible ELF/shared-object validation on current mainline, a working on-device application update flow with rollback, support for reusing packaged shared libraries, documentation for sim and hardware validation, and reviewable upstream changes integrated with Makefile and CMake workflows.
<p>In recent years, “implicit 3D representations” have been attracting attention as an appropriate 3D expression for deep neural networks. At CVPR 2019, three dif- ferent solutions using this 3D representation were proposed (Occupancy Networks, DeepSDF, and IM-NET), and they show that this representation is better in some as- pects(e.g., accuracy, memory efficiency) than voxel grids or meshes, which are usually used in previous studies. Nowadays, there are two major deep learning frameworks: TensorFlow and Py- Torch. Each of them has some libraries such as TensorFlow Graphics or Kaolin, which help users to use 3D models in neural networks easily. Unfortunately, none of the three works mentioned above are implemented in TensorFlow Graphics, whereas Kaolin provides their code. This is not a good situation for TensorFlow users who want to try using "implicit 3D representations" in their experiments or projects. In this proposal, I propose a project to implement these three works in Tensorflow Graphics. I will also work on equipping some 3D geometry utilities to support users to handle the three baselines on the same interface.</p>
<p>DiffEqSensitivity.jl is a Julia package within the SciML ecosystem for computing (adjoint) sensitivities for various kinds of differential equations in a highly performant manner within a unified user interface. In this project, we aim at implementing continuous adjoint sensitivity methods for hybrid differential equations as well as shadowing methods for chaotic dynamical systems in the DiffEqSensitivity.jl package. Moreover, we will add a symbolic preprocessing option based on the ModelingToolkit.jl package to build highly optimized adjoint differential equation code in DiffEqSensitivity.jl. Finally, we plan to refactor DiffEqSensitivity.jl to make use of a new package, namely AbstractDifferentiation.jl, such that users may straightforwardly select the best performing automatic-differentiation backend in each part of their program without the need to adapt their code substantially.</p> <p>Possible fields of application for these tools range from model discovery with explicit dosing times in pharmacology, over accurate gradient estimates for chaotic fluid dynamics, to the control of open quantum systems.</p>
BookBrainz still has a relatively small community and contains less entities than other comparable databases. Therefore we want to provide a way to import available collections of library records into the database while still ensuring that they meet BookBrainz’ high data quality standards. From a previous GSoC project in 2018, the database schema already contains additional tables set up for that purpose, where the imports will await a user’s approval before becoming a fully accepted entity in the database. The project will require processing very large data dumps (e.g. MARC records or JSON files) in a robust way and transforming entities from one database schema to the BookBrainz schema. Additionally the whole process should be repeatable without creating duplicate entries. My previous GSoC project from 2023 has achieved this for standalone entities, but it is not very useful to import works or editions without being able to link them to their authors, for example. So there is still a good amount of work ahead to adapt this process to import a full set of related entities and deal with all the edge cases.
As industrial manufacturing accelerates its digital transformation through advancements in robotics, adaptive production lines, and smart testing systems, cloud-edge collaboration has emerged as a critical enabler for deploying embodied intelligence in complex operational environments. Contemporary industry requirements for embodied intelligence now extend beyond basic task execution to encompass multimodal perception and decision integration, dynamic environment adaptation, and distributed device orchestration. Existing benchmarking frameworks exhibit limitations in evaluating scenario-specific embodied attributes inherent to industrial settings. This initiative leverages the KubeEdge-Ianvs collaborative AI framework, integrating domain-specific test datasets, simulation environments, and quantitative metrics to establish a certified industrial-grade evaluation infrastructure for embodied intelligence systems. Expected Outcome: 1. Develop an industrial-grade embodied intelligence dataset through systematic classification and reorganization of existing resources/ examples 2. Deploy baseline algorithms and introduce metrics to establish performance benchmarks within KubeEdge-Ianvs
<p>Living systematic reviews have been proposed as a new approach to deal with the main problem of traditional systematic reviews. Systematic reviews can rapidly become out of date and updating is often delayed by the publication model.</p> <p>metawRite package have the objective to rapid updating of systematic reviews as new research becomes available, while also ensuring a transparent process and reproducible review.</p> <p>Three main tasks are needed to have an effective living systematic review; the ability to produce dynamic reports, availability online with an interface that enables end users to understand the data and the ability to efficiently update the review(and any meta-analysis) with new research seamless. metawRite will cover these three task integrated in a friendly web based environment for the final user.</p> <p>One of the main functions of this package will display a modern web-based application for update a living systematic review. To do that I plan to combine three powerfull R packages, shiny, knitr and ggplot2 .</p>
<p>Vorbis as a lossy audio codec provides excellent quality audio compared to other codecs at a similar bit rate, including MP3, and even comparable to AAC. High-quality audio combined with a fully open license should be valued by the open source community as an alternative to codecs that require a patent license for encoding or payments in order to legally stream.</p> <p>However, FFmpeg’s native vorbis encoder is currently labeled as experimental or broken due to its relatively poor performance and audio quality in comparison to competing audio codecs. It serves only to output a valid bitstream, with little concern about quality. This project aims to improve the general performance of the encoder, bringing it closer to a viable status. This will be done by integrating features of the existing AAC psychoacoustic model system within the native Vorbis encoder and adapting the FFmpeg Vorbis decoder to utilize that new information. Finally, additional documentation and test coverage will be added to enable long-term development in this area.</p>
<p>The primary objective of this project is to analyze the role of mainframe platforms in achieving digital sovereignty and operational resilience within enterprise architectures. The mentee will evaluate the strengths and weaknesses of on-premise mainframe infrastructure compared to public cloud environments, specifically focusing on data residency, regulatory compliance (such as DORA), and resilience models. The research will explore how mainframes address critical sovereignty challenges, including third-party concentration risk and data access governance, where distributed systems may introduce vulnerabilities. Additionally, the project will assess the technical limitations of mainframe-centric strategies, such as skills dependencies, and provide a strategic analysis of hybrid sovereign architectures.</p><p><br></p><p>Deliverables: A whitepaper to be published under the modernization working group banner.</p><p><br></p><p><strong>Applications closing date: 25-Apr-2026</strong></p><p><br></p><p><strong>Mentee confirmation: 10-May-2026</strong></p>
<p>Pocket Science Lab is a complete and powerful tool that can really power a regular science lab. It can work as Oscilloscope and Logical Analyzer. PSLab can generate desirable waveforms, voltages, currents and pwm signals, it can read resistance, capacitance and data from sensors (wireless/analog/digital). Its features have opened new horizons for engineers, teachers and students. As we live in 21st century, where Smartphones are possessed by the majority of people, it is important that the features of PSLab can be accessed via Smartphone. This proposal aims towards the development of an Android application for PSLab experiments. The proposal proposes an entire model contributing towards the development of the project. It starts with discussing various aspects of the project and defining its goals. Then it focusses on project implementation, the software lifecycle to be used and the modules in which the system can be divided into. Subsequently, it states application’s ideology and thoroughly explains each module (along with wireframes). It also presents a timeline of the entire project. The proposal concludes by discussing development and testing phase of the project.</p>
<p>The Taverna suite is consist of Taverna Engine, which handles both Taverna Workbench and the Taverna Server. Taverna Server is used to execute remote workflows and workbench provides GUI for create, run, import workflows.This concept allow scientists who has limited knowledge in computing, limited technical resources to do complex process on Data. Also Taverna engine can be run on standard Windows, Linux or Mac OS. At the moment Taverna workflows are constructed using SCULF2 API. So, Tavera Engine can’t execute other workflows other than Taverna workflows. This is where CWL comes in ,The Common Workflow Language (CWL). CWL allows one description of a workflow to be run on multiple different platforms as long as they all support CWL. " CWL builds on technologies such as JSONLD and Avro for data modeling and Docker for portable runtime environments ". The project mission is to bring CWL UI support in the Workbench which is a graphical way to build workflows and one of the steps in Add Common Workflow Language support to Taverna.</p>
<p>ChemClipse is a solution developed under The Eclipse Foundation to explore and visualize analytical data of Chemistry from more than one source applications. R programming language is mostly used by researchers for statistical and analytical applications. It leverages systems not only by providing statistical tests and models but also by strengthening with methods to manage and manipulate large amount of information. ChemClipse will benefit with this capability.</p> <p>During Google Summer of Code-2017, I intend to systematically introduce fundamental R capabilities to ChemClipse. My plan is to utilize RJ open source library and provide interactive console as well as scripting windows for users. I am going to additionally support it by building a prototype of GUI, for users with non-programming skills, which will be based on the pipeline laid out for interactive console window.</p> <p>The users will require programming skills in R language for using interactive console. This risk can be mitigated by extending console functionalities and enhancing prototype of GUI. Thus, users not having programming skills can also benefit.</p>
Open Food Facts Explorer is a next-generation frontend for exploring and contributing to the Open Food Facts ecosystem, built with modern technologies like SvelteKit. It aims to achieve full feature parity with the main platform by keeping it modern, scalable and contributor-friendly. This project aims to enhance Explorer by adding multi-platform support (Open Food Facts, Open Beauty Facts, Open Pet Food Facts, and Open Products Facts), improving product addition and editing workflows, integrating Open Prices and moderator only features and tools like Nutri-Patrol directly into the app. It will also introduce enhanced search capabilities and Robotoff-powered features to assist contributors. From a technical perspective, the focus will be on decoupling the frontend from the backend by strengthening the SDK usage, and leveraging reusable web components where appropriate while maintaining SvelteKit’s SSR benefits. “Transform openfoodfacts-explorer from a food-only prototype into a complete, multi-platform contribution tool that achieves feature parity with the main OFF website"
<p>In the last several years, deep learning became widely used in various different fields, ranging from object detection and face recognition to speech recognition and machine translation. The common thing between all these areas is that they are working on structured data - data that can be represented in a table, such as images with a well defined number of rows and columns. An example of unstructured data are graphs - a collections of nodes connected with edges. Although graphs can also be represented through adjacency matrix, this representation implies ordinality of the nodes. Furthermore, once trained on a graph represented this way, a neural network could be used only on the graphs of the same sizes. This was fixed with the introduction of graph neural networks.</p> <p>In this project, I will construct models based on graph neural networks and apply them to graphs found in the Reactome database. Reactome is a graph database of pathways of several organisms. The goal is to create a Python client able to compute the vector representations of the pathways, which may be used in downstream analysis such as predicting whether a pathway is involved within a certain disease.</p>
<p>I would like to work on improving background subtraction algorithm. In particular, I want to implement algorithm from the following article: <a href="http://www.cv-foundation.org/openaccess/content_cvpr_2016_workshops/w24/papers/Guo_Background_Subtraction_Using_CVPR_2016_paper.pdf" target="_blank">Background Subtraction using Local SVD Binary Pattern, Guo et al. (2016)</a></p> <p>It is based on LSBP feature descriptors and achieves state-of-the-art performance on the CDnet 2012 dataset. LSBP descriptors are particularly good in regions with illumination variation, noise and shadows. So, this algorithm has better performance in this kind of regions.</p> <p>In addition, this algorithm has another pleasant trait: after extraction of LSBP descriptors it processes frames pixel-wise (i.e. independently). This is the common trait among all the consensus-based background models. It means, that implementation can leverage the parallelism inside OpenCV. Furthermore, implementation could be fully parallelized and implemented on both CPU/GPU. Thus, this algorithm will be fast enough for the realtime processing even on mobile devices.</p>
<p>There is a deep learning module in OpenCV named DNN, which has a set of built-in layers and functionality for loading serialized networks models from different frameworks. There are many matrix multiples and add operation in many layers, such as the convolution layer. These operations can be optimized by SIMD and some convolution operation optimizations have been optimized in<a href="https://github.com/opencv/opencv/blob/master/modules/dnn/src/layers/layers_common.simd.hpp" target="_blank"> layers_common.simd.hpp</a> with AVX intrinsics. RISC-V vector extension also have many useful intrinsics to optimize deep learning kernels. I believe that we can imitate it to implement the optimized code with RVV Intrinsic by following way: First, Create a code block wrapped by the macro "CV_RVV" which can help us distinguishing the implementation of optimize OpenCV DNN on different platforms. Secondly, use RISC-V V intrinsics instead of AVX intrinsics but keep the function unchanged. Finally, test the code using QEMU.</p>
<p>Carrier synchronization in standard, mass-market GNSS receivers typically utilizes well understood locked-loop architectures. The performance obtained with such architectures is sufficient for benign propagation scenarios, but typically deliver poor performance under harsh propagation conditions. Code and carrier tracking, as well as joint code/carrier synchronization, can be formulated as an estimation problem which can be solved using Bayesian filtering methods. It has been shown in the GNSS literature that KF-based synchronization solutions can be used to overcome the performance limitations of standard approaches, offering implicitly adaptive filter bandwidth, and opening up the possibility of using nonlinear models to avoid certain limitations associated with the use of code, phase or frequency discriminators. In this contribution, we will leverage powerful nonlinear tracking algorithms, including cubature, unscented, and sigma point Kalman filters in order to produce realizations of carrier and joint code-carrier tracking blocks which promise to be more effective and adaptable in challenging GNSS environments when compared to traditional architectures.</p>
<p>Everyone wants to focus on improving their model performance, but a significant portion of their time is used up while searching for a proper dataset and then preparing it. TensorFlow datasets (TFDS) library is a huge relief for all the researchers trying to avoid data preprocessing.<br> So to make TFDS users’ life even more comfortable, I would like to propose the following features and improvements:</p> <ul> <li>Load datasets without reading the dataset generation code. (This will unlock backward compatibility and cross-language support for reading datasets)</li> <li>Implement folder-based methods to prepare custom Image datasets, Translation datasets, etc.</li> <li>Add a command-line tool to simplify the procedure to add new datasets in TFDS.</li> <li>Simplify the procedure to add datasets outside TFDS.</li> <li>Improve Windows compatibility.</li> <li>Improve the TFDS catalog.</li> <li>More user and contributor friendly documentation and tutorials.</li> <li>Solve tfds issues :-)</li> </ul>
The objective of the "Physics-based Sonar Simulation for Underwater Robotics" project is twofold: firstly, to transition to ROS2 and integrate with the latest Gazebo simulation environment, and secondly, to substantially enhance the plugin's functionality. A key goal is to diversify the range of sonar models within the simulation, thereby significantly extending its relevance and efficacy across a variety of underwater conditions and operational scenarios. This strategic expansion aims to provide a more versatile and comprehensive tool for simulating dynamic underwater environments, facilitating advanced research and development in autonomous underwater navigation and obstacle avoidance. Deliverables: 1. A fully migrated and functional physics-based sonar simulation plugin within the ROS 2 framework and the LTS version of the new Gazebo. 2. Enhanced performance and capabilities of the plugin to support higher refresh rates. 3. A comprehensive suite of tests and validations conducted in simulated environments, ensuring reliability before real-world deployments. 4. Extended functionality through the inclusion of additional sonar types, catering to a broader spectrum of research and operational needs.
Drupal's AI Content Strategy module does a great job telling editors what to write, but the moment it hands off, editors are left staring at a blank field with no actual help getting words on the page. This project builds a content generation assistant that sits directly inside Drupal field widgets, letting editors prompt an AI model and insert the result without ever leaving the editing interface. The module will be built around a clean plugin architecture that supports OpenAI GPT, Anthropic Claude, Hugging Face, and Google Gemini as interchangeable providers, with a React-based prompt dialog giving editors control over tone, length, and style. A REST endpoint will bridge the frontend widget to the backend provider system, and the module will include generation history tracking, rate limiting, and cost cap controls for site administrators. By the end of GSoC, the project will ship as a fully documented, publicly released Drupal module that closes the gap between strategic content recommendations and actual content creation.
<p>Storage is one of the main limiting factors to the recording of information from proton-proton collision events at the Large Hadron Collider (LHC), at CERN in Geneva. Hence, the ATLAS experiment at the LHC uses a trigger system, which selects and transfers interesting events to the data storage system. However, if those events are buried in large backgrounds and difficult to identify as a signal by the trigger system, they will also be discarded together with the background. To alleviate this problem, various compression algorithms are proposed to reduce the size of the data that is recorded. One of those algorithms is an autoencoder (AE) network that tries to implement an approximation to the identity, f(x) = x. Its goal is to create a lower-dimensional representation of the input data in a latent space. Then using this latent representation the model can reconstruct its input. Moreover, two advanced families of AEs are proposed, the Variational and the Adversarial AE. Next, various combinations of different compression algorithms are tested to compress and construct accurate representations of the input data. Finally, the studied AEs are implemented for anomaly detection.</p>
<p>The nature of distributed systems and their fault model is chaotic: replicas can fail temporarily or just stop working, and the complexity of adding defensive code against all possible scenarios is unrealistic if complex applications are to be designed. Despite several techniques that try to mitigate this issue (e.g. RPC), a more opaque abstraction layer is needed in order to avoid long implementation times as well as incorrect behaviour driven from faulty or careless code. I believe that providing a programming-language-level abstraction layer is an effective approach to significantly reduce the amount of time focused on how systems can fail, rather than how they should work. Such an approach should mainly consist in supplying systems programmers with basic functionalities in the following areas: node distribution, data replication and partition tolerance. Being in an early adoption stage, Lasp needs applications that can serve as proof of concept in order to increase awareness of its capabilities as well as user adoption. This proposal is about the implementation of a real world application based on previous work published recently at PaPoC'17 - an EuroSys workshop.</p>
<p>The Alliance of Genome Resources is a consortium of 6 model organism databases (MODs) and the Gene Ontology (GO) Consortium whose goal is to provide an integrated view of their data to all biologists, clinicians and other interested parties. Alliance of Genomic Resources now has several services which require authentication and authorization. The aim of this project is to implement a single service to maintain authentication and authorization across the project as the number of services increase.</p> <p>This authentication and authorization service will allow users to sign in using Oauth2 (Google, Microsoft, etc...). For the Alliance members, JWT access tokens and refresh tokens with specific user information will be generated, which could be used for authentication of Literature service at this time and other services in the future (File Management System(FMS), Public Website, Annotation Database, Curation Interface).</p> <p>The implementation of this project consists three parts: Authentication and authorization backend service for Alliance members, User Login for non-Alliance members and A Frontend interface for user login and shows user statistics.</p>
<p>OpenCV.js exposes JavaScript API of dnn module that allows web apps to do deep learning model inference in web browsers. This capability enables web developers to create intelligent usages, like image classification, object detection, segmentation and style transfer, with rich web contents, like images, videos and camera streams. Although OpenCV.js dnn module is using WebAssembly with multi-threading and SIMD128 optimization, there still exists a huge gap between OpenCV.js and OpenCV native. This is because OpenCV native can leverage more advanced hardware features and even ML-specific accelerators via native instructions and ML APIs. Web Neural Network API (WebNN) is an emerging web standard that allows web apps to access the AI hardware accelerations. It can get near-native performance on a variety of Intel-based platforms through the open source module oneDNN. Thus, in this proposal, I illustrate my idea about how to integrate WebNN into OpenCV.js and accelerate dnn module.</p>
The project looks at how to facilitate the process of Telecom roaming agreements drafting and negotiation. The project looks at first constructing a library of drafting articles with a set of variables that will be extracted from the available templates and previous roaming agreements using Natural Language Processing process NLP. The second part will be towards translating the drafting and negotiation process as a chaincode on Blockchain to digitalize the process and provide a maintainable and actionable copy of the agreement. The project focus on how to be able to digitalize the legal process and make it automated through smart contracts to be more efficient and less error prone. The expected outcome of this project, would include: 1. Building a drafting library and an NPL model that will include a digital version of the most relevant and important articles in a the GSMA AA.12, AA.13 and AA-14 roaming agreement templates. 2. Developing a chaincode that will map all steps of the drafting and negotiation business processes for roaming agreements. 3. Building a PoC for UI to handle the agreement construction 4. Writing a solution document summarizing the implementation details