An Analysis of Deep Object Detectors For Diver Detection

Abstract

With the end goal of selecting and using diver detection models to support human-robot collaboration capabilities such as diver following, we thoroughly analyze a large set of deep neural networks for diver detection. We begin by producing a dataset of approximately 105,000 annotated images of divers sourced from videos -- one of the largest and most varied diver detection datasets ever created. Using this dataset, we train a variety of state-of-the-art deep neural networks for object detection, including SSD with Mobilenet, Faster R-CNN, and YOLO. Along with these single-frame detectors, we also train networks designed for detection of objects in a video stream, using temporal information as well as single-frame image information. We evaluate these networks on typical accuracy and efficiency metrics, as well as on the temporal stability of their detections. Finally, we analyze the failures of these detectors, pointing out the most common scenarios of failure. Based on our results, we recommend SSDs or Tiny-YOLOv4 for real-time applications on robots and recommend further investigation of video object detection methods.

14 pages, submitted for ICRA 21

References in corpus (11)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• YOLOv4: Optimal Speed and Accuracy of Object Detection
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• R-FCN: Object Detection via Region-based Fully Convolutional Networks
• A Survey of Deep Learning-based Object Detection
• Exploring the Landscape of Spatial Robustness
• Measuring Robustness to Natural Distribution Shifts in Image Classification
• Looking Fast and Slow: Memory-Guided Mobile Video Object Detection
• Using Videos to Evaluate Image Model Robustness
• On The Stability of Video Detection and Tracking
• Shift Equivariance in Object Detection