PulseAugur / Brief
EN
LIVE 11:13:37

Brief

last 24h
[2/2] 223 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. TOOL · arXiv cs.CV English(EN) ·

    Classifying galaxies in the Galaxy10 DECals dataset using Inception and Residual CNNs

    Researchers have evaluated the performance of ResNet101 and InceptionV4 convolutional neural networks for classifying galaxy images. Both models achieved approximately 90% accuracy on the Galaxy10 DECals dataset, demonstrating their suitability for future astronomical surveys. The study found ResNet101 to be slightly superior to InceptionV4 in terms of performance metrics. AI

    IMPACT Demonstrates the effectiveness of existing CNN architectures for astronomical image classification, potentially streamlining future research.

  2. TOOL · arXiv cs.LG English(EN) ·

    Balancing Real and Synthetic Data for CNN-based Masonry Crack Detection

    Researchers have developed a method to improve the accuracy of detecting cracks in masonry using convolutional neural networks (CNNs). They found that training CNNs with a combination of synthetic and real-world crack images significantly enhances performance. Specifically, using synthetic data with just 20% real data achieved results comparable to, and in some cases better than, using only real data. AI

    IMPACT This research could lead to more efficient and accurate structural health monitoring systems by reducing the need for extensive real-world data collection.