Guy Levesque’s Post

📣 Road to #iccv2023, paper 2 of 3. 📣 Shyam Nandan Rai, #ELLIS PhD student at Politecnico di Torino in the VANDAL lab, had his paper titled "Unmasking Anomalies in Road-Scene Segmentation" accepted. The paper proposes a new paradigm for anomaly segmentation, based on the recent trend of universal segmentation architecture. The idea proposed in the paper is to use a mask-based architecture (Mask2Former) to lift the prediction of anomalies at the level of masks. This idea is based on the observation that the classic per-pixel reasoning is prone to produce noisy predictions, thus leading to a high number of false positives and poorly localized anomalies. On the other hand, a mask based approach encourages objectness and thus can capture anomalies as whole entities. Our final Mask2Anomaly architecture also includes a series of improvements that boost performance, achieving outstanding results on a variety of datasets. Kudos to Shyam and thanks to Fabio Cermelli and Dario Fontanel for their precious contribution! 📗 ArXiv: https://lnkd.in/dNHPgjxS 💻 Code (coming soon): https://lnkd.in/dn9DG6_N 🤖 VANDAL lab: http://vandal.polito.it/ SMIYC results should appear in the leaderboard soon (we are just waiting for the request to be processed) A shoutout also to the paper "RbA: Segmenting Unknown Regions Rejected by All" from Fatma Güney's group, which ha also been accepted to ICCV2023 and it also uses a mask-classification approach. 🔥 This reinforces the idea of using a mask-based reasoning to detect o.o.d. elements in segmentation tasks. ELLIS - European Laboratory for Learning and Intelligent Systems

Unitos was privileged to have organised an industrial visit to the students of Milam Engineering college. The students of Artificial intelligence and computer science department benefited out of this program. The program highlighted the application of AI in data deliverable. CSE students were apprised of software application used for post processing of drone data. #artificialintelligence #engineering #computerscience #dronetechnology #postprocessing

Paper Acceptances 1. Water Q2 Impact Factor: 3.4 Dear Prof. Shaikh Congratulations on the acceptance of your manuscript, and thank you for submitting your work to Water: Manuscript ID: water-2654726 Type of manuscript: Article Title: Forecasting Floods of Bangladesh Using Machine Learning and Deep Learning Algorithms 2. CMES Q2 Impact Factor: 2.4 Dear Authors We are pleased to inform you that your submission ID: 43921 titled "Social Media-Based Surveillance Systems for Health Informatics Using Machine and Deep Learning Techniques: A Comprehensive Review and Open Challenges" to CMES-Computer Modeling in Engineering & Sciences has been officially accepted. https://lnkd.in/eHxg5wwf

🌟 Exciting Update: Day 150 of 200 Days of ML Journey! 🚀 Today marks a significant milestone in my 200 Days of ML journey as I dived deep into the realm of Generative Models and Applications. Here's a glimpse of what I covered: 🔍 Explored the significance of generative models in computer vision, understanding their impact on various industries. 🤖 Studied Generative Adversarial Networks (GANs) and their incredible ability to generate realistic images. 🖼️ Delved into Pix2Pix, a powerful conditional GAN architecture used for image-to-image translation tasks. 🌈 Explored Pix2Pix's applications in image colorization, style transfer, and super-resolution tasks. I also got hands-on experience by implementing a Pix2Pix model and experimenting with different datasets and hyperparameters to optimize its performance. #MachineLearning #GenerativeModels #GANs #Pix2Pix #ImageTranslation #ComputerVision #200DaysOfML GitHub Repo: https://lnkd.in/gsMMNzYX Can't wait to share more insights and progress with you all on this amazing journey! Keep learning, and keep growing! 💪🎉

Had a fantastic presentation today, on one of the problems from my research project: Traffic Trajectory Prediction in a Heterogeneous Environment. It was an internal presentation among fellow researchers from the School of Computer, Data and Mathematical Sciences. My talk, titled "Graphs in the Trajectory Prediction: Beyond Distance," focused on the significance of graphs in trajectory prediction and the limitations of current Graph-based methods. I also presented my proposed solution for improvement! Looking forward to finalizing the research article for publication soon. #Research #traffictrajectoryprediction #graphs #autonomousvehicles #ADAS #TrajectoryPrediction #soCDMS #WSU

🎓🖼️ Exciting News! 🖼️🎓 I'm thrilled to share that I've successfully completed the Image Processing Certificate from MathWorks, AICTE! 📜 This course offered by MathWorks. Image processing is an essential part of many scientific and engineering applications, including computer vision, medical imaging, and remote sensing. This course consists of several modules that cover the fundamental concepts of image processing. Some of the key topics covered in the course include: • Introduction to Image Processing: This module has covered the basics of digital images, including image representation, pixel values, and color spaces. • Image Processing with MATLAB: In this module, I have learnt how to use MATLAB to load, manipulate, and display images. • Image Enhancement: This module has covered techniques for improving the appearance and quality of images, including filtering, contrast adjustment, and sharpening. • Image Restoration: This module has covered methods for restoring images that have been degraded by noise, blur, or other sources of distortion. • Segmentation and Feature Extraction: This module has covered techniques for dividing an image into regions and extracting features, such as edges, corners, and textures. One of the unique features of the Image Processing Onramp course is that it has provided a hands-on learning experience.#imageprocessing #mathworks #aicteinternship #aicte

Here, I am sharing some of the accomplishments that i have done recently. I have enrolled in an industrial training on computer vision & image processing organized by #EdulakesSolutionLLP and learned many new thing related #open_cv2, image transformation and feature extraction. I have also completed one new research paper #Evaluation_of_Different_Kernels_for_Spatial_Filtering_Techniques_Under_Image_Processing and published it on #IJAEAST and received the publication certificate.

Thrilled to announce that I presented my research paper, "Measuring and Predicting Students Academic Performance using ML," at the prestigious Congress on Smart Computing Technologies (CSCT 2023) held on December 2-3, 2023! My research delves into the potential of machine learning to predict student academic performance, aiming to empower educators with data-driven insights for personalized support and improved learning outcomes. The paper explores various ML algorithms and analyzes their effectiveness in predicting student performance based on various factors. Presenting at CSCT 2023 provided an invaluable platform to connect with fellow researchers and discuss cutting-edge advancements in educational technology. #CSCT2023 #MachineLearning #EducationTechnology #Research #AcademicPerformance #AIforGood

Really excited to announce that, some months ago, the article "Machine Learning Techniques Applied to the Development of a Fall Risk Index for Older Adults" was published on the journal IEEE. I would like to thank Rodrigo Pérez-Rodríguez and Fundación para la Investigación Biomédica del Hospital Universitario de Getafe for the opportunity to contribute to this project. The article focuses on the fact that falls are a leading cause of unintentional trauma-related deaths worldwide, and a significant contributor to elderly dependence. The goal was to predict recurrent falls in the older population using machine learning techniques, with the aim of reducing the number of falls and their consequences. The link to the article can be found below: https://lnkd.in/dzk8gT8f