News
A novel approach to Open Set Domain Adaptation leverages unknown exploration, enhancing classification boundaries and improving model adaptability.
Fine-tuned Segment Anything Model (SAM) achieves state-of-the-art lung segmentation in X-ray images. Presented at CIARP 2024.
Our innovative TransferAttn framework enhances Vision Transformers for Unsupervised Domain Adaptation in videos, achieving groundbreaking results in action recognition.
LaSID participated in WebMedia'24, presenting two papers, with one earning Honorable Mention. The event, in its 30th edition, highlights innovations in multimedia and web technologies.
Guideline for intrinsic evaluation of item embeddings in recommender systems proposes innovative methods to assess qualitative properties.
Innovative study uses edge computing and deep learning for real-time plant disease detection, enhancing precision agriculture.
New machine learning system recommends personalized AML therapies, achieving high accuracy. Study funded by CAPES, CNPq, and FAPESP.
Revolutionary system encodes biomedical data to accelerate drug discovery for Acute Myeloid Leukemia. Supported by FAPESP grants.