Accurate recognition of multiple fish species is essential in marine ecology and fisheries. Precisely classifying and tracking these species enriches our comprehension of their movement patterns and empowers us to create precise maps of species-specific territories. Such profound insights are pivotal in conserving endangered species, promoting sustainable fishing practices, and preserving marine ecosystems’ overall health and equilibrium. To partially address these needs, we present a proposed model that combines YOLOv8 for object detection with ByteTrack for tracking. YOLOv8’s oriented bounding boxes help to improve object detection across angles, while ByteTrack’s robustness in various scenarios makes it ideal for real-time tracking. Experimental results using the SEAMAPD21 dataset show the model’s effectiveness, with YOLOv8n being the lightweight yet modestly accurate option, suitable for constrained environments. The study also identifies challenges in fish tracking, such as lighting variations and fish appearance changes, and proposes solutions for future research. Overall, the proposed model shows promising fish tracking and counting results, which is essential for monitoring marine life.
Fish species must be identified for stock assessments, ecosystem monitoring, production management, and the conservation of endangered species. Implementing algorithms for fish species detection in underwater settings like the Gulf of Mexico poses a formidable challenge. Active learning, a method that efficiently identifies informative samples for annotation while staying within a budget, has demonstrated its effectiveness in the context of object detection in recent times. In this study, we present an active detection model designed for fish species recognition in underwater environments. This model can be employed as an object detection system to effectively lower the expense associated with manual annotation. It uses epistemic uncertainty with Evidential Deep Learning (EDL) and proposes a novel module denoted as Model Evidence Head (MEH) for fish species detection in underwater environments. It employs Hierarchical Uncertainty Aggregation (HUA) to obtain the informativeness of an image. We conducted experiments using a fine-grained and extensive dataset of reef fish collected from the Gulf of Mexico, specifically the Southeast Area Monitoring and Assessment Program Dataset 2021 (SEAMAPD21). The experimental results demonstrate that an active detection framework achieves better detection performance on the SEAMAPD21 dataset demonstrating a favorable balance between performance and data efficiency for underwater fish species recognition.
The deep neural network has found widespread application in object detection due to its high accuracy. However, its performance typically depends on the availability of a substantial volume of accurately labeled data. Several active learning approaches have been proposed to reduce the labeling dependency based on the confidence of the detector. Nevertheless, these approaches tend to exhibit biases toward high-performing classes, resulting in datasets that do not adequately represent the testing data. In this study, we introduce a comprehensive framework for active learning that considers both the uncertainty and the robustness of the detector, ensuring superior performance across all classes. The robustness-based score for active learning is calculated using the consistency between an image and its augmented version. Additionally, we leverage pseudo-labeling to mitigate potential distribution drift and enhance model performance. To address the challenge of setting the pseudo-labeling threshold, we introduce an adaptive threshold mechanism. This adaptability is crucial, as a fixed threshold can negatively impact performance, particularly for low-performing classes or during the initial stages of training. For our experiment, we employ the Southeast Area Monitoring and Assessment Program Dataset 2021 (SEAMAPD21), comprising 130 fish species classes with 28,328 image samples. The results show that our model outperforms the state-of-the-art method and significantly reduces the annotation cost. Furthermore, we benchmark our model’s performance against a public dataset (PASCAL VOC07), showcasing its effectiveness in comparison to existing methods.
For autonomous driving, pedestrian and road signs detection are key elements. There is much existing literature available addressing this issue successfully. However, the autonomous system requires a large and diverse set of training samples and labeling in real-world environments. Manual annotation of these samples is somewhat challenging and time-consuming. In this paper, our goal is to get better detection accuracy with minimal training data. For this, we have employed the active learning algorithm. Active learning is a useful method that selects only the effective portion of the dataset for training and reduces annotation costs. Though it uses only a small amount of the training data, it provides a high detection accuracy. In this work, we have chosen the deep active learning model for object detection via the probabilistic model of Choi et al. and modified the depth scale of different layers in the backbone. As real-world data may contain noise, motion, or other disruptions, we modified the original model to obtain improved detection results. In this experiment, we create a customized dataset that contains pedestrians, road signs, traffic lights, and zebra (or pedestrian) crossings to deploy the active learning algorithm. The experimental results show that the active learning model can produce good detection outcomes by accurately detecting and classifying pedestrians, road signs, traffic light, and zebra (or pedestrian) crossings.
Video surveys are commonly used to monitor the abundance and distribution of managed species to support management. However, considerable effort, time, and cost are required for human review and automated fish species recognition provides an effective solution to remove the bottleneck of post-processing. Implementing fish species detection techniques for underwater imagery is a challenging task. In this work, we present the Multiple Instance Active-learning for Fish-species Recognition (MI-AFR), which is formulated as an object detection-based approach to perform localization and classification of fish species. It can select the most informative fish images from unlabeled sets by estimating the uncertainty of unlabeled images by using adversarial classifiers trained on labeled sets. Moreover, we have analyzed the improved performance of MI-AFR by considering different backbone networks as a trade-off between speed and accuracy. For experiments, we have used the fine-grained and large-scale reef fish dataset obtained from the Gulf of Mexico – the Southeast Area Monitoring and Assessment Program Dataset 2021 (SEAMAPD21). The experimental results illustrate that the superiority of the proposed method can establish a solid foundation for active learning in fish species recognition, especially with a small number of labeled sets.
Fish species recognition and detection are essential for fishery industries. Accurate and robust species classification and detection play a vital role in monitoring fish activities and identifying the distribution of a specific species, which is vital to know the endangered species. It is also essential for controlling production and overall ecosystem control and management. However, the role of current artificial intelligence technologies, such as deep learning, is limited in the ocean system compared to other areas like robotics and security. The major challenge in building a deep learning network is data availability, time, and cost of annotation and labeling. In this work, we build a semi-supervised deep-learning network to recognize fish species. The model is based on a student-teacher network where the teacher network generates pseudo-labels, and the student network is trained with the generated pseud-labels and the labeled data simultaneously. The student network updates the teacher network via an exponential moving average method. The model consists of a faster R-CNN with a feature pyramid network detector. The experimental result of the model on the challenging fish dataset shows a promising result for building semi-supervised object detection models.
Species recognition is an important aspect of video based surveys, which support stock assessments, inspecting the ecosystem, handling production management, and protecting endangered species. It is a challenging task to implement fish species detection algorithms in underwater environments. In this work, we introduce the YOLOv5 model for the recognition of fish species that can be implemented as an object detection model for analyzing multiple fishes in a single image. Moreover, we have modified the depth scale of different layers in the backbone of the YOLOv5 model to obtain improved results on fish species recognition. In addition, we have implemented a transformer block in the backbone network and introduced a class balance loss function to obtain enhanced performance. It can perform fish species recognition as an object detection approach by classifying each of the fish species in addition to localizing for the estimation of the position and size of the fish in an image. Experiments are conducted on the fine-grained and large-scale reef fish dataset that we have obtained from the Gulf of Mexico – the Southeast Area Monitoring and Assessment Program Dataset 2021 (SEAMAPD21). The experimental results demonstrate that an enhanced YOLOv5 model can yield better detection results in comparison to YOLOv5 for underwater fish species recognition.
Adaptive cruise control (ACC), a common feature in an autonomous vehicle, is intended to automatically adjust the vehicle speed and maintain a safe distance from its preceding vehicle to avoid a collision. The main challenge is to filter the sensor data accurately, and the control system can make a decision quickly. This paper proposed a control method for ACC using the Extended Kalman filter (EKF) and a Proportional Integral Derivative (PID) controller, which can estimate the acceleration or braking of the preceding vehicle by adjusting the speed of the following vehicle. The proposed control method is assessed under various PID parameters using a Genetic Algorithm (GA) to optimize the ACC system using four loss metrics: (1) throttle loss, which accounts for fuel usage, and is proportional to the throttle setting; (2)ride quality, which is penalized by an excessive jerk (the first derivative of acceleration); (3) a distance penalty, which measures how far compared to the safe distance
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.