• Corpus ID: 245668933

Underwater Object Classification and Detection: first results and open challenges

@article{Jesus2022UnderwaterOC,
  title={Underwater Object Classification and Detection: first results and open challenges},
  author={Andre Jesus and Claudio Zito and Claudio Tortorici and Eloy Roura and Giulia De Masi},
  journal={ArXiv},
  year={2022},
  volume={abs/2201.00977}
}
This work reviews the problem of object detection in underwater environments. We analyse and quantify the shortcomings of conventional state-of-the-art (SOTA) algorithms in the computer vision community when applied to this challenging environment, as well as providing insights and general guidelines for future research efforts. First, we assessed if pretraining with the conventional ImageNet is beneficial when the object detector needs to be applied to environments that may be characterised by… 

Figures and Tables from this paper

References

SHOWING 1-10 OF 30 REFERENCES
Research on underwater object recognition based on YOLOv3
TLDR
This paper mainly locates and classifies the images of seacucumber, scallop, seaurchin and matches the real-time detection requirements in this experiment using two algorithms that are widely used at present to texperiment with underwater image dataset.
Toward a Generic Diver-Following Algorithm: Balancing Robustness and Efficiency in Deep Visual Detection
TLDR
This letter designs an architecturally simple convolutional neural network based diver detection model that is much faster than the state-of-the-art deep models yet provides comparable detection performances.
YOLO Nano Underwater: A Fast and Compact Object Detector for Embedded Device
TLDR
This paper proposes a fast and compact object detector for several marine products, such as scallop, starfish, echinus and holothurian, modified based on YOLO Nano architecture to reduce the inference time and achieves a competitive performance on edge computing devices, like NVIDIA Jetson Nano.
Fast Underwater Image Enhancement for Improved Visual Perception
TLDR
The proposed conditional generative adversarial network-based model is suitable for real-time preprocessing in the autonomy pipeline by visually-guided underwater robots and provides improved performances of standard models for underwater object detection, human pose estimation, and saliency prediction.
Reveal of Domain Effect: How Visual Restoration Contributes to Object Detection in Aquatic Scenes
TLDR
The relation of quality-diverse data domain to detection performance is investigated and visual restoration is beneficial to detection in the wild by reducing the domain shift between training data and real-world scenes.
WQT and DG-YOLO: towards domain generalization in underwater object detection
TLDR
A data augmentation method Water Quality Transfer (WQT) is proposed to increase domain diversity of the original small dataset and DG-YOLO is proposed, which consists of three parts: YOLOv3, DIM and IRM penalty, for mining the semantic information from data generated by WQT.
Detection of Marine Animals in a New Underwater Dataset with Varying Visibility
TLDR
This paper presents a new publicly available underwater dataset with annotated image sequences of fish, crabs, and starfish captured in brackish water with varying visibility, and is the first annotated underwater image dataset captured in temperateBrackish waters.
Enhancing Underwater Imagery Using Generative Adversarial Networks
TLDR
A method to improve the quality of visual underwater scenes using Generative Adversarial Networks (GANs), with the goal of improving input to vision-driven behaviors further down the autonomy pipeline is proposed.
CADDY Underwater Stereo-Vision Dataset for Human-Robot Interaction (HRI) in the Context of Diver Activities
TLDR
A novel underwater dataset collected from several field trials within the EU FP7 project “Cognitive autonomous diving buddy (CADDY)”, where an Autonomous Underwater Vehicle was used to interact with divers and monitor their activities, is presented.
UDD: An Underwater Open-sea Farm Object Detection Dataset for Underwater Robot Picking
TLDR
A lightweight network (UnderwaterNet) is designed to address the problems that detecting small objects from cloudy underwater pictures and meeting the efficiency requirements in robots and the effectiveness of the proposed UnderwaterNet is verified.
...
1
2
3
...