Faster R-CNN based Fish Detector for Smart Aquaculture System

Rosales, Marife A.; Palconit, Maria Gemel B.; Almero, Vincent Jan D.; Concepcion, Ronnie; Magsumbol, Jo-Ann V.; Sybingco, Edwin; Dadios, Elmer P.

doi:10.1109/hnicem54116.2021.9732042

Cited by 10 publications

(7 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is made up of two parts, namely, the loss computation and the backbone feature extraction network. The loss function is modified to Arcface Loss (AFL), and the backbone network that is used for feature extraction is improved based on DenseNet-169 [ 92 ]. The following is a list of the most important improvements made to DenseNet-169:…”

Section: Methodsmentioning

confidence: 99%

“…If there is no overlap between the values that are expected and the values that are found on the ground, then the IoU will be equal to zero, and the GIoU will finally be equal to one. As a consequence of this, we concluded that we should refer to this function as 1 –GIoU loss [ 92 ]. When there is a greater distance between the bounding boxes that are predicted and those that represent the ground truth, the ranges of IoU and GIoU are [0, 1] and [–1, 1], respectively.…”

Section: Methodsmentioning

confidence: 99%

“…DenseNet-169. This research utilizes and improves the DenseNet-169 network because, in comparison to DenseNet-121 and ResNet, the performance of DenseNet-169 is much higher [90][91][92][93]. Cardinality is a new hyperparameter that has been added to DenseNet with version 169 [92].…”

Section: Proposed Modelmentioning

confidence: 99%

“…This research utilizes and improves the DenseNet-169 network because, in comparison to DenseNet-121 and ResNet, the performance of DenseNet-169 is much higher [90][91][92][93]. Cardinality is a new hyperparameter that has been added to DenseNet with version 169 [92]. It was stated in the cited work [93] that increasing cardinality is a more effective strategy for attaining accuracy than expanding either the depth or the breadth of the considerations [94].…”

Section: Proposed Modelmentioning

confidence: 99%

“…Batch Normalization (BAN). The input data are converted into a normal distribution via BAN [92], which has a mean value of 0 and a variance value of 1. Using this approach, the network's capacity for generalization, initial learning rate, and convergence speed may all be improved.…”

Section: Plos Onementioning

confidence: 99%

See 4 more Smart Citations

Multi-classification deep neural networks for identification of fish species using camera captured images

et al. 2023

View full text Add to dashboard Cite

Regular monitoring of the number of various fish species in a variety of habitats is essential for marine conservation efforts and marine biology research. To address the shortcomings of existing manual underwater video fish sampling methods, a plethora of computer-based techniques are proposed. However, there is no perfect approach for the automated identification and categorizing of fish species. This is primarily due to the difficulties inherent in capturing underwater videos, such as ambient changes in luminance, fish camouflage, dynamic environments, watercolor, poor resolution, shape variation of moving fish, and tiny differences between certain fish species. This study has proposed a novel Fish Detection Network (FD_Net) for the detection of nine different types of fish species using a camera-captured image that is based on the improved YOLOv7 algorithm by exchanging Darknet53 for MobileNetv3 and depthwise separable convolution for 3 x 3 filter size in the augmented feature extraction network bottleneck attention module (BNAM). The mean average precision (mAP) is 14.29% higher than it was in the initial version of YOLOv7. The network that is utilized in the method for the extraction of features is an improved version of DenseNet-169, and the loss function is an Arcface Loss. Widening the receptive field and improving the capability of feature extraction are achieved by incorporating dilated convolution into the dense block, removing the max-pooling layer from the trunk, and incorporating the BNAM into the dense block of the DenseNet-169 neural network. The results of several experiments comparisons and ablation experiments demonstrate that our proposed FD_Net has a higher detection mAP than YOLOv3, YOLOv3-TL, YOLOv3-BL, YOLOv4, YOLOv5, Faster-RCNN, and the most recent YOLOv7 model, and is more accurate for target fish species detection tasks in complex environments.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%