Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The authors introduce SteelDS, a high-resolution video dataset of E40 steel and copper scrap designed for training and evaluating machine learning models for object detection and instance segmentation in the context of automated industrial sorting. The dataset contains 24,297 labeled frames with pixel-wise segmentation masks and material classes for 396 steel and 101 copper objects. This benchmark enables the development of algorithms to identify and remove copper impurities from steel scrap, potentially improving the efficiency and effectiveness of recycling processes.
Automating the tedious and costly removal of copper contaminants from steel scrap just got a boost: a new high-resolution video dataset is available for training better machine learning models.
This dataset provides high-resolution, annotated video sequences of shredded E40-grade steel and copper scrap on a conveyor belt. Captured in a controlled laboratory environment, the data reflects the industrial post-magnetic sorting stage, where manual intervention is typically required to remove copper contaminants. The dataset comprises 24,297 labeled frames across five subsets, featuring 396 steel and 101 copper objects categorized by size. It supports the development of machine learning models for material classification, object detection, and instance segmentation. Variations in object spacing and density are included to simulate realistic industrial sorting conditions. Ground truth annotations include pixel-wise segmentation masks and material classes. This dataset serves as a benchmark for evaluating automated sorting algorithms aiming to identify copper impurities within complex, heterogeneous steel scrap streams.