Yiwen Ma

Yiwen Ma

Auto-generated profile, based on publicly available information

Yiwen Ma

Beijing Normal Univ, Amur Tiger & Amur Leopard Monitoring & Res Ctr, Key Lab Conservat Ecol Northeast Tiger & Leopard N, Natl Forestry & Grassland Adm,Minist Educ,Key Lab, Beijing 100875, Peoples R China

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 August 2022 in Animals
Verified
Journal Article
Open access
PDF file
MTMengyu Tan
Mengyu Tan
WCWentao Chao
Wentao Chao
JCJo-Ku Cheng
Jo-Ku Cheng
MZMo Zhou
Mo Zhou
YMYiwen Ma
Yiwen Ma
Limin FengLimin Feng
Limin Feng
Camera traps are widely used in wildlife surveys and biodiversity monitoring. Depending on its triggering mechanism, a large number of images or videos are sometimes accumulated. Some literature has proposed the application of deep learning techniques to automatically identify wildlife in camera trap imagery, which can significantly reduce manual work and speed up analysis processes. However, there are few studies validating and comparing the applicability of different models for object detection in real field monitoring scenarios. In this study, we firstly constructed a wildlife image dataset of the Northeast Tiger and Leopard National Park (NTLNP dataset). Furthermore, we evaluated the recognition performance of three currently mainstream object detection architectures and compared the performance of training models on day and night data separately versus together. In this experiment, we selected YOLOv5 series models (anchor-based one-stage), Cascade R-CNN under feature extractor HRNet32 (anchor-based two-stage), and FCOS under feature extractors ResNet50 and ResNet101 (anchor-free one-stage). The experimental results showed that performance of the object detection models of the day-night joint training is satisfying. Specifically, the average result of our models was 0.98 mAP (mean average precision) in the animal image detection and 88% accuracy in the animal video classification. One-stage YOLOv5m achieved the best recognition accuracy. With the help of AI technology, ecologists can extract information from masses of imagery potentially quickly and efficiently, saving much time.

29 October 2024 in Global Ecology and Conservation
Verified
Journal Article
Open access
JLJiawei Li
Jiawei Li
DLDongrui Li
Dongrui Li
YFYanwen Fu
Yanwen Fu
YMYiwen Ma
Yiwen Ma
JGJianping Ge
Jianping Ge
Limin FengLimin Feng
Limin Feng
The sett is a crucial resource for badgers, serving various functions and significantly influencing their spatial ecology. The selection of sett sites is also a complex behavior. While there have been extensive ecological studies on the European badger (Meles meles), little is known about the Asian badger (Meles leucurus), particularly in China, making it difficult to evaluate their current survival status and establish conservation strategies. Therefore, we conducted a pioneering ecological study on the Asian badger, focusing on local population density and habitat selection within the Northeast Tiger and Leopard National Park, NE China. First, we surveyed badger setts in the study area and set camera traps to capture videos of badgers and estimate the number of individuals. The estimated badger density based on the main sett distribution (1.40 sett /km2) and group size (average of 3.01 ± 1.82 individuals per sett) was 4.21 ± 2.55 badgers /km2. Second, we set up random points in the study area and measured 12 habitat factors of 67 random point quadrats and 67 badger sett quadrats to establish which factors affect the habitat selection of badgers. Statistical tests, random forest algorithm, and generalized linear model were used to explain the sett selection criteria of Asian badgers. Statistical tests indicated a highly significant difference in slope (p < 0.01) and a significant difference in tree distance and aspect (p < 0.05) between sett quadrats and random point quadrats. After correction for multiple testing according to Holm, we found that differences in tree distance and aspect lost their statistical significance. However, the random forest algorithm and generalized linear model showed that the selection of sett location was influenced by tree distance, slope, and aspect. We conclude that badgers exhibit a preference for setts located near trees, steep slopes, and orientations facing north and northwest (316–45°). This is the first assessment of the abundance and sett selection of the Asian badger at the local population scale from Chinese nature reserves, providing new insights into the population ecology and habitat selection of the Asian badger in China, and assisting in the development of relevant conservation strategies and management.