PipeMFL-240K: A Large-scale Dataset and Benchmark for Object Detection in Pipeline Magnetic Flux Leakage Imaging
Researchers have introduced PipeMFL-240K, a large-scale dataset and benchmark designed to advance object detection in pipeline magnetic flux leakage (MFL) imaging. The dataset addresses the lack of public resources for MFL interpretation, which has hindered deep learning model development and comparison. PipeMFL-240K features over 249,000 images with more than 200,000 bounding-box annotations, presenting challenges such as a long-tailed distribution of 12 categories, tiny objects, and significant intra-class variability. This new resource aims to accelerate algorithmic innovation and enable reproducible research in pipeline integrity assessment. AI
IMPACT Provides a crucial benchmark for developing more reliable AI models in industrial safety and environmental protection.