Three critical factors affecting automated image species recognition performance for camera traps

Ecological camera traps are increasingly used by wildlife biologists to unobtrusively monitor an ecosystems animal population. However, manual inspection of the images produced is expensive, laborious, and time‐consuming. The success of deep learning systems using camera trap images has been previously explored in preliminary stages. These studies, however, are lacking in their practicality. They are primarily focused on extremely large datasets, often millions of images, and there is little to no focus on performance when tasked with species identification in new locations not seen during training. Our goal was to test the capabilities of deep learning systems trained on camera trap images using modestly sized training data, compare performance when considering unseen background locations, and quantify the gradient of lower bound performance to provide a guideline of data requirements in correspondence to performance expectations. We use a dataset provided by Parks Canada containing 47,279 images collected from 36 unique geographic locations across multiple environments. Images represent 55 animal species and human activity with high‐class imbalance. We trained, tested, and compared the capabilities of six deep learning computer vision networks using transfer learning and image augmentation: DenseNet201, Inception‐ResNet‐V3, InceptionV3, NASNetMobile, MobileNetV2, and Xception. We compare overall performance on “trained” locations where DenseNet201 performed best with 95.6% top‐1 accuracy showing promise for deep learning methods for smaller scale research efforts. Using trained locations, classifications with <500 images had low and highly variable recall of 0.750 ± 0.329, while classifications with over 1,000 images had a high and stable recall of 0.971 ± 0.0137. Models tasked with classifying species from untrained locations were less accurate, with DenseNet201 performing best with 68.7% top‐1 accuracy. Finally, we provide an open repository where ecologists can insert their image data to train and test custom species detection models for their desired ecological domain.     

浙江九龙山国家级自然保护区黑熊活动监测 及其华东地区潜在分布区预测

2017至2020年,在浙江九龙山国家级自然保护区内及周边网格化布设63台Ltl-6210MC红外相机,对区内的黑熊(Ursus thibetanus)及其同域物种进行调查。监测期间,5台红外相机共拍摄到9张黑熊活动照片和6次黑熊活动视频,提取到7次黑熊有效照片。根据本次监测到的黑熊位点与文献记录,共确定华东地区黑熊出现位点14个。基于黑熊栖息地特征选择土地利用和7个生境因子为预测背景,利用物种分布模型预测黑熊在华东地区的潜在分布区。为提高预测的精准度,采用了biomod 2软件包中的10种模型算法,并用真实技能统计值(TSS)和曲线下面积值(AUC)来评估这10种算法,只有当TSS值超过0.8且AUC值大于0.9时,才使用该模型算法预测物种的潜在分布区。结果表明:(1)推测保护区内现存2或3只黑熊;(2)最冷月份最低温(Bio6)和最湿季节降水量(Bio16)是限制黑熊分布的主要环境因子;(3)黑熊在华东地区存在3个主要潜在分布区,即浙–赣潜在分布区、浙–赣–皖潜在分布区和浙–皖潜在分布区,适生区面积约为317km~2。综上所述,这些结果可为加强华东地区黑熊的保护提供一定的资料。     

Near Infrared Optical Projection Tomography for Assessments of β-cell Mass Distribution in Diabetes Research

By adapting OPT to include the capability of imaging in the near infrared (NIR) spectrum, we here illustrate the possibility to image larger bodies of pancreatic tissue, such as the rat pancreas, and to increase the number of channels (cell types) that may be studied in a single specimen. We further describe the implementation of a number of computational tools that provide: 1/ accurate positioning of a specimen''s (in our case the pancreas) centre of mass (COM) at the axis of rotation (AR)²; 2/ improved algorithms for post-alignment tuning which prevents geometric distortions during the tomographic reconstruction² and 3/ a protocol for intensity equalization to increase signal to noise ratios in OPT-based BCM determinations³. In addition, we describe a sample holder that minimizes the risk for unintentional movements of the specimen during image acquisition. Together, these protocols enable assessments of BCM distribution and other features, to be performed throughout the volume of intact pancreata or other organs (e.g. in studies of islet transplantation), with a resolution down to the level of individual islets of Langerhans.     

Factors influencing the diurnal behaviour of white rhino (Ceratotherium simum) in Matobo National Park,Zimbabwe

Activity periods and diurnal behaviour of southern white rhino (Ceratotherium simum simum) were studied in Matobo National Park, Zimbabwe, from August 2016 to July 2017 using camera trapping and focal group sampling, respectively. Camera trap data showed that diurnal and nocturnal movements of white rhino varied significantly with moon phase, with more frequent nocturnal photographic captures during full- and third-quarter moon phases than during new and crescent moon phases. Permutational MANOVA and canonical correspondence analysis of focal group observation data were used to test the effects of season, time of day, habitat type, distance to roads, moon phase, presence/absence of other mammals, age/sex category, individual identity and herd composition on daytime behaviour. Weather conditions accounted for the majority of variation in rhino behaviour, with more time spent feeding during cloudy and rainy weather than sunny weather. Time of day and moon phase also accounted for high variation in behaviour, followed by habitat attributes and age/sex category. Variance partitioning by sets of explanatory variables showed that weather conditions had the greatest explanatory power in the variation of rhino behaviour, followed by temporal attributes, habitat attributes and age/sex category. Our findings provide insights for white rhino population management strategies aimed at increasing the effectiveness of security patrols, minimising disturbance and improving sightings for tourists.     

卧龙保护区三种中型食肉动物的生态位差异

2014至2019年,利用红外相机对卧龙保护区的赤狐（Vulpes vulpes）、黄喉貂（Martes flavigula）和豹猫（Prionailurus bengalensis）进行了监测,并分析比较了三者的栖息地分布和日活动节律。结果显示,赤狐栖息地主要分布于保护区西部高海拔地区,与黄喉貂和豹猫栖息地重叠较少;豹猫和黄喉貂的栖息地均集中分布于皮条河以南,且相互重叠部分达56.7%;在日活动节律方面,赤狐和豹猫表现出夜行性,黄喉貂表现出昼行性。本研究表明,三种食肉动物均通过栖息地或日活动节律的差异产生了生态位分离,从而实现了在卧龙保护区内的相互共存。     

Radiated temperature from thermal imaging is related to feed consumption,growth rate and feed efficiency in grower pigs

Individual feed consumption and animal weight were continuously recorded in grower pigs using an automated feeding system. Infrared images were recorded each time a pig entered the feeding system and infrared thermography provided radiated thermal measurements of the dorsal surface of each animal. Feed was withdrawn and the animals fasted for a period of 24 h three times during the growth of the animals at body weights of approximately 35, 65 and 105 kg. There was a significant reduction of 0.28 °C in the maximum surface temperature (Tmax), and 0.48 °C in the average surface temperature (Tmean) during the periods of fasting. Maximum and average pig temperatures exhibited negative correlations to feed consumption and growth variables. There were negative correlations of residual feed intake (RFI) to Tmax and Tmean radiated temperatures. There were positive correlations of residual gain (RG) and residual intake and gain (RIG) with Tmax and Tmean. The Tmax and Tmean temperature responses to fasting were negatively associated with feed consumption and growth variables. Absolute temperature and temperature response variables were positively associated with RFI and negatively associated with residual intake and gain (RIG). These findings provide support for the concept of radiated heat losses as a measure of metabolic activity and a predictor of growth performance.     

Human disturbance and the activity patterns and temporal overlap of tapirs and jaguars in reserves of NW Belize

Human disturbance from tourism and other non-consumptive activities in protected areas may be stressful to wildlife. Animals may move away in space or time to avoid human interaction. For species of particular conservation concern, such as Baird's tapirs (Tapirus bairdii) and jaguars (Panthera onca), a better understanding of how they respond to different levels and types of disturbance is needed in order to manage human visitation to parks in ways that minimize negative outcomes for wildlife. We describe the overlap in activity patterns of tapirs, jaguars, and humans at logged and unlogged sites and at places with low versus high human visitation using camera survey data from protected areas of NW Belize, 2013–2016. Tapirs were nocturnal in all study sites, with > 80% of all tapir detections occurring between 1900 hr and 0500 hr. Their activity patterns were not different in unlogged versus logged sites and did not change with increased human traffic. Jaguars were cathemeral across sites but had more nocturnal activity at the site with the most human impact. Activity pattern overlap between tapirs and jaguars did not differ significantly between logged and unlogged sites, nor between areas with low and high human activity. Human traffic increased from 2013 to 2016 at most of the study locations. In conclusion, this camera trap dataset suggests that non-consumptive human disturbance does not alter the activity patterns of tapirs and jaguars in protected areas lacking hunting pressure.     

利用红外线相机监测地表水对广西弄岗国家级自然保护区兽类分布的影响

广西弄岗国家级自然保护区属岩溶地貌,表土层薄,储水度低,地表水奇缺,动物可能因水源而改变其活动范围和规律。2006年3月—2007年6月和2010年10月—2011年5月,该实验利用红外线相机在兽类通往水源地之间,即兽径上(如足迹和粪便)进行兽类调查。共拍摄到兽类5目12科17属19种,包括国家一级保护动物黑叶猴(Trachypithecus francoisi)和熊猴(Macaca assamensis)。物种累积曲线95%置信区间重迭,相似性分析和非度量多维标度显示两种微生境出没物种的差异较小,然而,差异物种占整体记录的42%,可见岩溶地区水资源极其珍贵。中、短期保护建议包括加强水源点的巡护、防止偷猎、限制滥抽地下水,以及考虑整合呈岛屿状的陇呼、弄岗及陇山地区,以加强保护大面积生境的连贯性,并加速恢复山峰间已遭破坏的圆洼谷地的原生树林。     

Changing use of camera traps in mammalian field research: habitats,taxa and study types

相似文献   

Quantum chemical insight on vibration spectra of silica systems

A set of silicate ions and corresponding lithium salts have been quantum chemically (QC) simulated in a “free molecule” approach. The infrared (IR), inelastic neutron scattering (INS), and Raman spectra have been simulated and fitted to the experimentally registered ones. The complete assignment of the vibrational bands along with the intensities and potential energy distribution has been performed. The applicability of the traditionally used quasimolecule Si–O–Si model to the interpretation of bands near 440–480 cm^? 1 and so-called “Boson” peak near 50 cm^? 1 has been critically discussed.