Underwater Object Detection Based on Improved EfficientDet

Jia, Jiaqi; Fu, Min; Li, Xuefeng; Zheng, Bing

doi:10.3390/rs14184487

Cited by 33 publications

(19 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Deep learning techniques are the most used algorithms when detecting underwater objects, and these algorithms do not work the same in air as in underwater environments due to the environmental properties mentioned before. Works like [27] proposed a method for improving object detection in underwater environments based on an improved Ef-ficientDet. That method modified the structure of the neural network and the results showed that the mean average precision (mAP) reached 92.82%.…”

Section: Related Workmentioning

confidence: 99%

Inspection Operations and Hole Detection in Fish Net Cages through a Hybrid Underwater Intervention System Using Deep Learning Techniques

López-Barajas,

Sanz,

Marín-Prades

et al. 2023

JMSE

View full text Add to dashboard Cite

Net inspection in fish-farm cages is a daily task for divers. This task represents a high cost for fish farms and is a high-risk activity for human operators. The total inspection surface can be more than 1500 m2, which means that this activity is time-consuming. Taking into account the severe restrictions for human operators in such hostile underwater conditions, this activity represents a significant area for improvement. A platform for net inspection is proposed in this work. This platform includes a surface vehicle, a ground control station, and an underwater vehicle (BlueROV2 heavy) which incorporates artificial intelligence, trajectory control procedures, and the necessary communications. In this platform, computer vision was integrated, involving a convolutional neural network trained to predict the distance between the net and the robot. Additionally, an object detection algorithm was developed to recognize holes in the net. Furthermore, a simulation environment was established to evaluate the inspection trajectory algorithms. Tests were also conducted to evaluate how underwater wireless communications perform in this underwater scenario. Experimental results about the hole detection, net distance estimation, and the inspection trajectories demonstrated robustness, usability, and viability of the proposed methodology. The experimental validation took place in the CIRTESU tank, which has dimensions of 12 × 8 × 5 m, at Universitat Jaume I.

show abstract

Section: Related Workmentioning

confidence: 99%

Inspection Operations and Hole Detection in Fish Net Cages through a Hybrid Underwater Intervention System Using Deep Learning Techniques

López-Barajas,

Sanz,

Marín-Prades

et al. 2023

JMSE

View full text Add to dashboard Cite

show abstract

“…The model's architecture proposed in 9 is based on the MobileNetV2 network 10 and the 4 th version of YOLO 11 and is able to achieve an approximate average Mean Average Precision (mAP) of 86% over two public image dataset with an inference speed exceeding 44 Frames Per Second (FPS). The authors in 12 developed an improved version of EfficientDet 13 and compared it to Faster RCNN, 14 YOLOv4 11 and other detectors on two underwater datasets containing images of underwater animals that are collected using ROVs and other resources. Another group of researchers 15 compared the performance of Faster RCNN, 14 YOLOv3, 16 YOLOv4 11 RetinaNet 17 and SSD 18 detectors on the RUIE dataset 19 and concluded that different detectors offers different trade-offs and that YOLOv3 is the best performing detector on this dataset.…”

Section: Related Workmentioning

confidence: 99%

Multi-class plant type detection in great lakes region using remotely operated vehicle and deep learning

Saleem

Awad

Sidike

et al. 2023

Pattern Recognition and Tracking XXXIV

View full text Add to dashboard Cite

Underwater plants can alter the biodiversity and function of aquatic ecosystems, leading to significant positive or negative environmental impacts. In the Great Lakes region, many efforts have been made to early identify and map underwater plants for cost-effective monitoring, management, and containment of potentially invasive species. However, traditional patrolling and visual inspection methods are time-consuming and labor-intensive. Recent advances in sensors, computing platforms, and machine learning algorithms enabled unprecedented achievements in various applications related to marine biodiversity and aquatic ecosystem. In this work, we use an ROV to collect a new underwater plant detection dataset comprising 414 underwater images of three main categories of aquatic plants, namely Leafy, Bushy, and Tapey at different lakes in the Upper Peninsula, Michigan, USA. Then we present a comparative analysis of the performance of common object detectors, including YOLOv8, Faster R-CNN, and RetinaNet on our dataset. The acquired results demonstrate the potential of such pre-trained detectors in detecting underwater plants in noisy images that are acquired by ROVs and building fully automated plant detection, mapping, and management systems.

show abstract

“…However, it is known that DWConv can lead to a loss of information. To solve this problem, input features are expanded before applying DWConv [29]. This expansion step mitigates the potential loss of information.…”

Section: • Efficientnet B2mentioning

confidence: 99%

Optimizing Handwritten Arabic Character Recognition: Feature Extraction, Concatenation, and PSO-Based Feature Selection.

Rouabhi,

Belbachir,

Tlemsani

2023

Preprint

View full text Add to dashboard Cite

Recognizing and writing Arabic characters correctly is a difficult task, especially for non-native speakers. With the growing trend toward digital teaching and distance learning, there is a need for efficient and accurate automatic recognition of Arabic characters. The work investigates three distinct approaches to address this challenge. Firstly, features are extracted from handwritten Arabic character images using two pre-trained models: EfficientNet B2 and DenseNet 201. These extracted features are then employed for classification utilizing various classifiers. Secondly, the features obtained from both models are concatenated to enhance classification performance by leveraging the unique representations learned by each model. Lastly, the Particle Swarm Optimization (PSO) algorithm is integrated into the feature selection process, identifying the most relevant features that contribute to improved classification accuracy. The best test accuracies obtained were 85.94% in the first experiment, 88.73% in the second, and 90.20% in the third.

show abstract

Underwater Object Detection Based on Improved EfficientDet

Cited by 33 publications

References 53 publications

Inspection Operations and Hole Detection in Fish Net Cages through a Hybrid Underwater Intervention System Using Deep Learning Techniques

Inspection Operations and Hole Detection in Fish Net Cages through a Hybrid Underwater Intervention System Using Deep Learning Techniques

Multi-class plant type detection in great lakes region using remotely operated vehicle and deep learning

Optimizing Handwritten Arabic Character Recognition: Feature Extraction, Concatenation, and PSO-Based Feature Selection.

Contact Info

Product

Resources

About