Detecting rocks in challenging mining environments using convolutional neural networks and ellipses as an alternative to bounding boxes

作者：

Highlights：

• A detector named Rocky-CenterNet is proposed that detects rocks as ellipses.

• Four rock detectors are compared respect to performance and speed.

• Rocky-CenterNet runs 4x faster than Mask R-CNN.

• Rocky-CenterNet is a suitable alternative for mining applications.

摘要

•A detector named Rocky-CenterNet is proposed that detects rocks as ellipses.•Four rock detectors are compared respect to performance and speed.•Rocky-CenterNet runs 4x faster than Mask R-CNN.•Rocky-CenterNet is a suitable alternative for mining applications.

论文关键词：Object detection,Rock detection,Convolutional neural networks,Deep learning

论文评审过程：Received 1 February 2021, Revised 7 January 2022, Accepted 8 January 2022, Available online 17 January 2022, Version of Record 24 January 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.116537