Automatic counting of birds in a bird deterrence field trial

相似文献   

Automatically detecting and tracking free‐ranging Japanese macaques in video recordings with deep learning and particle filters

Recently, automated observation systems for animals using artificial intelligence have been proposed. In the wild, animals are difficult to detect and track automatically because of lamination and occlusions. Our study proposes a new approach to automatically detect and track wild Japanese macaques (Macaca fuscata) using deep learning and a particle filter algorithm. Macaque likelihood is derived through deep learning and used as an observation model in a particle filter to predict the macaques’ position and size in an image. By using deep learning as an observation model, it is possible to simplify the observation model and improve the accuracy of the classifier. We investigated whether the algorithm could find body regions of macaques in video recordings of free‐ranging groups at Katsuyama, Japan to evaluate our model. Experimental results showed that our method with deep learning as an observation model had higher tracking accuracy than a method that uses a support vector machine. More generally, our study will help researchers to develop automatic observation systems for animals in the wild.     

基于机器视觉和深度学习的稻纵卷叶螟性诱智能监测系统

【目的】为减轻基层测报人员工作量,提高稻纵卷叶螟Cnaphalocrocis medinalis性诱测报的准确率和实时性,实现监测数据可追溯,建立了基于机器视觉的稻纵卷叶螟性诱智能监测系统。【方法】稻纵卷叶螟性诱智能监测系统包括基于机器视觉的智能性诱捕器、基于深度学习的稻纵卷叶螟检测模型、系统Web前端和服务器端。利用工业相机、光源和Android平板搭建了智能性诱捕器的机器视觉系统;建立了基于改进的YOLOv3和DBTNet-101双层网络的稻纵卷叶螟检测模型;利用HTML, CSS, JavaScript和Vue搭建系统Web前端展示稻纵卷叶螟检测与计数结果;使用Django框架搭建服务器端,对来自智能性诱捕器通过4G网络上传的图像进行接收与结果反馈;采用MySQL数据库保存图像和模型检测结果等信息。【结果】基于机器视觉的稻纵卷叶螟性诱智能监测系统利用智能性诱捕器自动定期上传稻纵卷叶螟图像至服务器,部署在服务器上的目标检测模型对稻纵卷叶螟成虫进行实时自动检测,精确率和召回率分别达97.6%和98.6%;用户可通过Web前端查看稻纵卷叶螟检测结果图。【结论】基于机器视觉的稻纵卷叶螟性...     

Comparisons among survey methodologies to test for abundance and size of a highly targeted fish species

Three sampling methods for estimating abundance and size of blue cod Parapercis colias were compared inside and outside Kapiti Marine Reserve, New Zealand (40° 49′ 31·77^′′ S; 174° 55′ 02·87^′′ E). Two baited methods, baited underwater video (BUV) and experimental angling (EA), were more efficient and had lower levels of estimate variation than diver‐based underwater visual census (UVC). The BUV and EA recorded more fish and of greater size ranges than UVC, and also had fewer zero count replicates. The BUV and EA methodologies revealed highly significant differences in abundance and size of fish between sites (reserve v. non‐reserve), whereas UVC revealed no such differences. These results indicate that BUV is likely to be the most accurate, cost‐effective and easy to use methodology for the surveying of carnivorous temperate reef fishes for future monitoring. It is noted, however, that new data acquired using the BUV methodology may need to be compared over a calibration period to data acquired using the UVC methodology to ensure that historical data sets derived from UVC still have validity and application for future monitoring activity.     

Larval dispersal and movement patterns of coral reef fishes,and implications for marine reserve network design

Well‐designed and effectively managed networks of marine reserves can be effective tools for both fisheries management and biodiversity conservation. Connectivity, the demographic linking of local populations through the dispersal of individuals as larvae, juveniles or adults, is a key ecological factor to consider in marine reserve design, since it has important implications for the persistence of metapopulations and their recovery from disturbance. For marine reserves to protect biodiversity and enhance populations of species in fished areas, they must be able to sustain focal species (particularly fishery species) within their boundaries, and be spaced such that they can function as mutually replenishing networks whilst providing recruitment subsidies to fished areas. Thus the configuration (size, spacing and location) of individual reserves within a network should be informed by larval dispersal and movement patterns of the species for which protection is required. In the past, empirical data regarding larval dispersal and movement patterns of adults and juveniles of many tropical marine species have been unavailable or inaccessible to practitioners responsible for marine reserve design. Recent empirical studies using new technologies have also provided fresh insights into movement patterns of many species and redefined our understanding of connectivity among populations through larval dispersal. Our review of movement patterns of 34 families (210 species) of coral reef fishes demonstrates that movement patterns (home ranges, ontogenetic shifts and spawning migrations) vary among and within species, and are influenced by a range of factors (e.g. size, sex, behaviour, density, habitat characteristics, season, tide and time of day). Some species move <0.1–0.5 km (e.g. damselfishes, butterflyfishes and angelfishes), <0.5–3 km (e.g. most parrotfishes, goatfishes and surgeonfishes) or 3–10 km (e.g. large parrotfishes and wrasses), while others move tens to hundreds (e.g. some groupers, emperors, snappers and jacks) or thousands of kilometres (e.g. some sharks and tuna). Larval dispersal distances tend to be <5–15 km, and self‐recruitment is common. Synthesising this information allows us, for the first time, to provide species, specific advice on the size, spacing and location of marine reserves in tropical marine ecosystems to maximise benefits for conservation and fisheries management for a range of taxa. We recommend that: (i) marine reserves should be more than twice the size of the home range of focal species (in all directions), thus marine reserves of various sizes will be required depending on which species require protection, how far they move, and if other effective protection is in place outside reserves; (ii) reserve spacing should be <15 km, with smaller reserves spaced more closely; and (iii) marine reserves should include habitats that are critical to the life history of focal species (e.g. home ranges, nursery grounds, migration corridors and spawning aggregations), and be located to accommodate movement patterns among these. We also provide practical advice for practitioners on how to use this information to design, evaluate and monitor the effectiveness of marine reserve networks within broader ecological, socioeconomic and management contexts.     

Robust and simplified machine learning identification of pitfall trap‐collected ground beetles at the continental scale

1. Insect populations are changing rapidly, and monitoring these changes is essential for understanding the causes and consequences of such shifts. However, large‐scale insect identification projects are time‐consuming and expensive when done solely by human identifiers. Machine learning offers a possible solution to help collect insect data quickly and efficiently.
2. Here, we outline a methodology for training classification models to identify pitfall trap‐collected insects from image data and then apply the method to identify ground beetles (Carabidae). All beetles were collected by the National Ecological Observatory Network (NEON), a continental scale ecological monitoring project with sites across the United States. We describe the procedures for image collection, image data extraction, data preparation, and model training, and compare the performance of five machine learning algorithms and two classification methods (hierarchical vs. single‐level) identifying ground beetles from the species to subfamily level. All models were trained using pre‐extracted feature vectors, not raw image data. Our methodology allows for data to be extracted from multiple individuals within the same image thus enhancing time efficiency, utilizes relatively simple models that allow for direct assessment of model performance, and can be performed on relatively small datasets.
3. The best performing algorithm, linear discriminant analysis (LDA), reached an accuracy of 84.6% at the species level when naively identifying species, which was further increased to >95% when classifications were limited by known local species pools. Model performance was negatively correlated with taxonomic specificity, with the LDA model reaching an accuracy of ~99% at the subfamily level. When classifying carabid species not included in the training dataset at higher taxonomic levels species, the models performed significantly better than if classifications were made randomly. We also observed greater performance when classifications were made using the hierarchical classification method compared to the single‐level classification method at higher taxonomic levels.
4. The general methodology outlined here serves as a proof‐of‐concept for classifying pitfall trap‐collected organisms using machine learning algorithms, and the image data extraction methodology may be used for nonmachine learning uses. We propose that integration of machine learning in large‐scale identification pipelines will increase efficiency and lead to a greater flow of insect macroecological data, with the potential to be expanded for use with other noninsect taxa.

     

A machine vision system to detect and count laying hens in battery cages

Manually counting hens in battery cages on large commercial poultry farms is a challenging task: time-consuming and often inaccurate. Therefore, the aim of this study was to develop a machine vision system that automatically counts the number of hens in battery cages. Automatically counting hens can help a regulatory agency or inspecting officer to estimate the number of living birds in a cage and, thus animal density, to ensure that they conform to government regulations or quality certification requirements. The test hen house was 87 m long, containing 37 battery cages stacked in 6-story high rows on both sides of the structure. Each cage housed 18 to 30 hens, for a total of approximately 11 000 laying hens. A feeder moves along the cages. A camera was installed on an arm connected to the feeder, which was specifically developed for this purpose. A wide-angle lens was used in order to frame an entire cage in the field of view. Detection and tracking algorithms were designed to detect hens in cages; the recorded videos were first processed using a convolutional neural network (CNN) object detection algorithm called Faster R-CNN, with an input of multi-angular view shifted images. After the initial detection, the hens’ relative location along the feeder was tracked and saved using a tracking algorithm. Information was added with every additional frame, as the camera arm moved along the cages. The algorithm count was compared with that made by a human observer (the ‘gold standard’). A validation dataset of about 2000 images achieved 89.6% accuracy at cage level, with a mean absolute error of 2.5 hens per cage. These results indicate that the model developed in this study is practicable for obtaining fairly good estimates of the number of laying hens in battery cages.     

A method for the automated long‐term monitoring of three‐spined stickleback Gasterosteus aculeatus shoal dynamics

This paper describes and evaluates a flexible, non‐invasive tagging system for the automated identification and long‐term monitoring of individual three‐spined sticklebacks Gasterosteus aculeatus. The system is based on barcoded tags, which can be reliably and robustly detected and decoded to provide information on an individual's identity and location. Because large numbers of fish can be individually tagged, it can be used to monitor individual‐ and group‐level dynamics within fish shoals.     

Behavioral signatures of structured feature detection during courtship in Drosophila

1. Download : Download high-res image (193KB)
2. Download : Download full-size image

     

Linking movement and reproductive history of brook trout to assess habitat connectivity in a heterogeneous stream network

相似文献   

Forest and connectivity loss drive changes in movement behavior of bird species

In a rapidly changing world, it is important to understand how environmental modifications by humans affect species behavior. This is not a simple task, since we need to deal with a multitude of species and the different external contexts that affect their behavior. Here, we investigate how interpatch short-distance movements of 73 common forest bird species can be predicted by forest cover and forest isolation. We modeled bird movement as a function of environmental covariates, species traits – body mass and feeding habit – and phylogenetic relationships using Joint Species Movement Models. We used field data collected in forest edges and open pastures of six 600 × 600 m plots in the Atlantic Forest biodiversity hotspot. We found that birds fly larger distances and visit more forest patches and remnant trees with decreasing forest cover. Increasing landscape isolation results in larger flight distances, and it increases the use of trees as stepping-stones for most species. Our results show that birds can adjust their behavior as a response to spatial modification in resource distribution and landscape connectivity. These adjusted behaviors can potentially contribute to ecosystem responses to habitat modification.     

The role of chromatic and achromatic signals for fruit detection by birds

Fruit color changes during ripening are typically viewed asan adaptation to increase signal efficacy to seed dispersers.Plants can increase signal efficacy by enhancing chromatic (wavelengthrelated) and/or achromatic (intensity related) contrasts betweenfruit and background. To assess how these contrasts determinethe detectability of fruit signals, we conducted 2 experimentswith free-flying crows (Corvus ossifragus) under seminaturalconditions in a 2025 m² aviary. Crows searched first for artificialred and black fruits and detected red fruits from a larger distance.Because artificial red fruits had higher chromatic and lowerachromatic contrasts against foliage than artificial black fruits,crows apparently prioritized chromatic contrasts. Thus, thecommon change in fruit color from red to black during ripeningdoes not increase signal efficacy to crows. In a second trial,crows searched for UV-reflecting and black blueberries (Vaccinummyrtillus) against backgrounds of foliage and sand. Againstfoliage, UV-reflecting berries had higher chromatic and achromaticcontrasts than black berries, and crows detected them from alarger distance. Against sand, UV-reflecting berries had lowachromatic contrasts and black berries low chromatic contrasts.Crows detected both fruit types equally, suggesting that theyused chromatic contrasts to detect UV-reflecting berries andachromatic contrasts to detect black berries. Birds prioritizedchromatic contrasts when searching for artificial red fruitsin foliage but not when searching for blueberries on sand. Wesuggest that the relative importance of chromatic and achromaticcontrasts is contingent on the chromatic and achromatic varianceof the background. Models of signal perception can be improvedby incorporating background-specific effects.     

基于F3Net显著性目标检测的蝴蝶图像前背景自动分割

[目的]具有复杂背景的蝴蝶图像前背景分割难度大.本研究旨在探索基于深度学习显著性目标检测的蝴蝶图像自动分割方法.[方法]应用DUTS-TR数据集训练F3Net显著性目标检测算法构建前背景预测模型,然后将模型用于具有复杂背景的蝴蝶图像数据集实现蝴蝶前背景自动分割.在此基础上,采用迁移学习方法,保持ResNet骨架不变,利...     

声景生态学数据分析与应用

声景生态学是研究景观中生物与非生物声音在多种时空尺度下的声学格局与过程,揭示声音与人类以及声音与自然之间关系的学科。基于声景生态学的研究内容,从声景元素解析、生物多样性评估及人类身心健康评价应用案例中,梳理了数据分析的前沿方法。结论表明,分析技术的发展,特别是人工智能技术的进步,使声景生态学的研究呈现从人工到机器、从单一特征计算到多维特征提取、从单学科研究到多学科联合分析的技术化发展趋势,不断拓展着声景生态学的研究深度与广度。同时这些分析技术的发展也急需优化和标准化,来提高方法的通用性和研究结果间的可比性。此外,需要融合生态学、计算机科学和心理学等交叉学科的理论和方法,进一步推动声景生态学数据分析技术方法体系的完善。     

“Micropersonality” traits and their implications for behavioral and movement ecology research

1. Many animal personality traits have implicit movement‐based definitions and can directly or indirectly influence ecological and evolutionary processes. It has therefore been proposed that animal movement studies could benefit from acknowledging and studying consistent interindividual differences (personality), and, conversely, animal personality studies could adopt a more quantitative representation of movement patterns.
2. Using high‐resolution tracking data of three‐spined stickleback fish (Gasterosteus aculeatus), we examined the repeatability of four movement parameters commonly used in the analysis of discrete time series movement data (time stationary, step length, turning angle, burst frequency) and four behavioral parameters commonly used in animal personality studies (distance travelled, space use, time in free water, and time near objects).
3. Fish showed repeatable interindividual differences in both movement and behavioral parameters when observed in a simple environment with two, three, or five shelters present. Moreover, individuals that spent less time stationary, took more direct paths, and less commonly burst travelled (movement parameters), were found to travel farther, explored more of the tank, and spent more time in open water (behavioral parameters).
4. Our case study indicates that the two approaches—quantifying movement and behavioral parameters—are broadly equivalent, and we suggest that movement parameters can be viewed as “micropersonality” traits that give rise to broad‐scale consistent interindividual differences in behavior. This finding has implications for both personality and movement ecology research areas. For example, the study of movement parameters may provide a robust way to analyze individual personalities in species that are difficult or impossible to study using standardized behavioral assays.

     

Methods for estimating detection efficiency and tracking acoustic tags with mobile transect surveys

Mobile transect surveys were conducted for 5 years in a deep fjord looking for weak acoustic tags, most of which were in dead fishes lying on the seabed. Detection efficiencies were quantified because inferences about fish presence were made from tag detection data but a considerable proportion of tags were not detected. Methods for assessing transect widths using cumulative probabilities of measured perpendicular distances from blind transect lines to triangulated tag locations are described. Three methods for estimating detection efficiencies are presented: fitting attenuation functions to perpendicular distance data, simultaneous use of two receiver systems and test tag surveys. Several factors that influenced these detection efficiencies and ranges are shown, including boat speed, depth of tag, tag type, different hydrophone–receiver systems and variation among individual hydrophones and receivers of the same model. The trade‐off between detection efficiency per unit area and area swept is discussed, which is largely based on boat speed during transects. Finally, several methods are suggested for increasing decoding efficiencies in mobile tracking studies when tags are heard but are difficult to decode.     

Automatic tracking of gold seed markers from CBCT image projections in lung and prostate radiotherapy

PurposeTo construct a method and software to track gold seed implants in prostate and lung patients undergoing radiotherapy using CBCT image projections.MethodsA mathematical model was developed in the MatLab (Mathworks, Natick, USA) environment which uses a combination of discreet cosine transforms and filtering to enhance several edge detection methods for identifying and tracking gold seed fiducial markers in images obtained from Varian (Varian Medical Systems, Palo Alto, USA) and Elekta (Kungstensgatan, Sweden) CBCT projections.ResultsOrgan motion was captured for 16 prostate patients and 1 lung patient.ConclusionImage enhancement and edge detection is capable of automatically tracking markers for up to 98% (Varian) and 79% (Elekta) of CBCT projections for prostate and lung markers however inclusion of excessive bony anatomy (LT and RT LAT) inhibit the ability of the model to accurate determine marker location.     

Using sutures to attach miniature tracking tags to small bats for multimonth movement and behavioral studies

Determining the detailed movements of individual animals often requires them to carry tracking devices, but tracking broad-scale movement of small bats (<30 g) has been limited by transmitter technology and long-term attachment methods. This limitation inhibits our understanding of bat dispersal and migration, particularly in the context of emerging conservation issues such as fatalities at wind turbines and diseases. We tested a novel method of attaching lightweight global positioning system (GPS) tags and geolocating data loggers to small bats. We used monofilament, synthetic, absorbable sutures to secure GPS tags and data loggers to the skin of anesthetized big brown bats (Eptesicus fuscus) in Colorado and hoary bats (Lasiurus cinereus) in California. GPS tags and data loggers were sutured to 17 bats in this study. Three tagged bats were recaptured 7 months after initial deployment, with tags still attached; none of these bats showed ill effects from the tag. No severe injuries were apparent upon recapture of 6 additional bats that carried tags up to 26 days after attachment; however, one of the bats exhibited skin chafing. Use of absorbable sutures to affix small tracking devices seems to be a safe, effective method for studying movements of bats over multiple months, although additional testing is warranted. This new attachment method has the potential to quickly advance our understanding of small bats, particularly as more sophisticated miniature tracking devices (e.g., satellite tags) become available.