Search

An accurate and rapid identification of species is one of the most important parts of freshwater fish pre-treatment. However, deep learning-based freshwater fish classification cannot quantitatively describe the feature parameters of fish, due to the complicated feature extraction. This study aims to propose a deep learning-based fish feature point detection and freshwater fish classification. Firstly, an image acquisition device was built to acquire the images of five freshwater fish species: bream, bighead carp, grass carp, silver carp, and common carp. Secondly, the data enhancement was performed on the original images. LabelMe software was also used to label 20 feature points of the fish body, in order to construct a freshwater fish feature point detection dataset. The feature point of freshwater fish was detected using an improved AlexNet feature point detection model with the convolutional neural networks (CNN). The improved AlexNet model also adjusted the network structure of the traditional model, in order to accommodate the feature point detection task by the modified convolutional kernel size, the removal of the local response normalization layer, the addition of batch normalization layer, and the replacement of the loss function. The Euclidean distance between feature points was extracted as the feature value using the coordinates of the feature points. Feature vectors were constructed. Fisher discriminant analysis was used to identify the species of freshwater fish. A series of ablation experiments were conducted on the improved AlexNet model to clarify the impact of different improvement schemes on model performance. The results showed that the improved schemes were beneficial to improve the training efficiency and performance of the model. The average value of normalized mean error (NME) of the improved AlexNet model on the test set was 0.0099, the failure rate (FR) at thresholds of 0.02 and 0.03 were 2.50% and 0.83%, and the average detection time was 0.037 s, indicating the better accuracy and error distribution. The number of parameters and floating-point operations per second (FLOPs) of the improved AlexNet model were smaller than those of the VGG16 model. The test set detection was also better than that of the VGG16 model, compared with different feature extraction models. Although the number of parameters and FLOPs of the improved AlexNet model was larger, the average NME and the FR were lower than that of the MobileNetV2 model. The recognition accuracy of 98.0% was achieved in the freshwater fish species recognition using the model and Fisher discriminant analysis for five freshwater fish species. The average recognition time of a single image was 0.368s. The feature point detection model can be expected to detect the feature points of five species of freshwater fish with high accuracy. The finding can also facilitate freshwater fish species identification, size detection, and fish segmentation. The freshwater fish species identification using feature point detection and Fisher's discriminant analysis can lay the foundation for the development of an automated freshwater fish classification device. [ABSTRACT FROM AUTHOR]

Copyright of Transactions of the Chinese Society of Agricultural Engineering is the property of Chinese Society of Agricultural Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Transactions of the Chinese Society of Agricultural Engineering is the property of Chinese Society of Agricultural Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Feeding has been one of the most cost sources in fish farming so far. Artificial and mechanized feeding has been the most two widely-applied approaches to fish feeding in China. However, both feeding cannot be cost saving, due mainly to the amount of feed always depends on the experience rather than the real hunger situation of fish. Once the feed is superfluous, it is likely to the water contamination in many cases. It is necessary to appropriately adjust the amount of feed, according to the timely feeding information. As a result, a more flexible feeding has been a high demand to control the feed amount, according to the changes in fish behavior, in order to effectively reduce costs for more benefits. Intelligent feeding has been partially realized, whether via the detected information from the feeding process through various sensors, or via the developed predicting models as a feeding guideline for the decision-making. However, there are still great challenges to realizing completely intelligent feeding. However, the predicted information depends mainly on the complex farming environment and uncertain fish behavior. In this review, the most technical applications were categorized in the fish intelligent feeding over the recent decades, including fish feeding behavior, visually intelligent feed detection, acoustics, sensor, and biological model technology. Several types of commonly-used fish feeders were determined, including the gravity drop, pneumatic conveying, centrifugal throwing, and hydraulic conveying types. A better performance was achieved to avoid the labor costs and feed wasting, if the feeder was precisely selected to fit well the actual breeding operation, according to the various fish types or breeding modes. Some aspects of the feeder still need to be improved for the goal of complete intelligence. More advanced techniques can be required to meet the future requirements of the feeder in the long-distance precise transmission for large-scale fish farming, specifically including average feeding and wide-range feeding. By contrast, an accurate and reliable feeding can be achieved for the feeder in the factory intensive farming mode, with emphasis on the small-size design and technical revolution. Furthermore, there is also some change from the timer and quantitative feeding to intelligent feeding in machines. Finally, a systematic investigation was made to summarize the national and international research on the specific application scenarios and basic components of intelligent feeding systems. It indicated that the signal transmission and control system were dominated in the most intelligent machines of fish feeding, where the input data was generally collected from the farming environment and fish behavior in practice. The document analysis on intelligent feeding demonstrated that the instant behavior of fish was usually identified to guide the feeding operation under the experimental condition in fish farming. Yet, there was still something to do better: 1) Some exogenous factors can pose some elusive impacts on the actual feeding operation, such as the environmental fluctuations or human activities; 2) The feeding decision-making models can be required to upgrade or modify for the wide universality suitable for the more types of scenarios and fish; 3) The computing equipment can be more powerful to promote more reliable data sampling. In conclusion, the main purpose of an intelligent feeding system can be to fit much better with the various feeding situations, whether different fish types or different breeding modes. As such, a much more accurate judgment can be achieved to take more variables into the comprehensive consideration. Therefore, it is a promising potential way to combine more available disciplines like Aquaculture, Engineering, and Computer Science, or incorporate more accessible information, such as images, sounds, growth patterns, and biological characteristics. The multidisciplinary or multimedia fusing entity can greatly contribute to the fish feeding system for fully complete intelligence in the near future. [ABSTRACT FROM AUTHOR]

Copyright of Transactions of the Chinese Society of Agricultural Engineering is the property of Chinese Society of Agricultural Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Sustainable agricultural mechanization has been one of the most important strategies for the decision-making and operations in the management of crop production. An accurate and rapid evaluation can greatly contribute to promoting the quality of sustainable agricultural mechanization in China. However, many complex and changing factors are confined to the sustainable agricultural mechanization in recent years. Generally, four dimensions can be maintained for agricultural mechanization: the social and economic benefits, the operation level, the impact on the resources and environment, and the level of the agricultural engineering industry. In this study, a systematic evaluation was made on the influencing factors in the sustainable agricultural mechanization in Hubei Province of China using the combination of Decision-Making Trial and Evaluation Laboratory (DEMATEL) and Interpretive Structure Modeling (ISM). An evaluation index system was constructed to determine 13 indicators of sustainable agricultural mechanization, four parts of which were pressure (including the positive and negative factors), state, influence, and response, according to the hierarchical connections. A comparison was also made on the specific data of the 13 secondary indicators in the study area from 2004 to 2020. The quantitative data showed that the fuzzy DEMATEL-ISM performed better to integrate the subjective and objective opinions of influence indicators, further to determine the main influencing factors of sustainable agricultural mechanization. A pressure-state-impact-response model was selected to assess the index system of t sustainable agricultural mechanization. It was found that the pressure factors were the most significant driving factors in agricultural mechanization, accounting for more than 40% of the weights in the evaluation index system, followed by the State, and the lowest were Impact and Response. The results show that the economic factors were the most important development factors, while, the environmental factors dramatically increased to be the influencing factors. Furthermore, the input-output, scientific and technological factors of agricultural mechanization were also comprehensively evaluated in the index system. More importantly, the evaluation result showed that the sustainable agricultural mechanization in the study area was positive and orderly developed as a whole over the past ten years. Among them, the positive driving factors of the pressure and state-related indicators increased significantly, such as the economics, the agricultural engineering industry, as well as agricultural science and technology. The maximum contribution rate was attributed to the economic benefits and capacity development in sustainable agriculture manufacturing. The pressure part was remained basically stable, indicating the impact of labor reduction on agricultural mechanization. The agricultural machinery cost factors were basically maintained a small range of fluctuations. The adverse factors of the energy environment posed a great effect on income, output growth, and agricultural mechanization. At the first time, the response part was much higher than the pressure and impact parts, indicating the ever-increasing investment in the control of sustainable development. Much attention can be paid to the concept of sustainable development for the decision-making on the increasing investment, further offering greater opportunities and challenges for the sustainable agricultural mechanization in the end, some recommendations were also proposed for the future development of agricultural mechanization. [ABSTRACT FROM AUTHOR]