Recognition of Rare Microfossils Using Transfer Learning and Deep Residual Networks
Corresponding AuthorYang, Xiaoguang( ; Zhang, Tao(
Wang, Bin1; Sun, Ruyue1; Yang, Xiaoguang2,3; Niu, Ben1; Zhang, Tao1; Zhao, Yuandi1; Zhang, Yuanhui1; Zhang, Yiheng1; Han, Jian2
Source PublicationBIOLOGY-BASEL
AbstractSimple Summary The collection of early Cambrian microfossils leads to the amassing of a pile of thousands of tiny tubes, grains and various fragments. Rare type of microfossils with high academic value are mingled with numerous ordinary fossils and the traditional way of manual selection is very inefficient. Many artificial intelligence (AI) technologies have already been applied in fossil image recognition, but current methods largely depend on a great number of fossil images to "train" the AI model. However, usually only a handful of samples are available for specific rare fossil taxa and these cannot provide enough photos for AI. In this study, we fine-tuned a new convolutional neural network, combining pre-trained models from a nature image database to solve the problem of the lack of training materials. Through many tests, this new model was proved valid. It presented relatively high accuracies in recognizing specific micro fossil taxa, while the required number of corresponding fossil images is very low. Various microfossils from the early Cambrian provide crucial clues for understanding the Cambrian explosion and the origin of animal phyla. However, specimens with important anatomical structures are extremely rare and the efficiency of retrieving such fossils by traditional manual selection under a microscope is quite low. Such a contradiction has hindered breakthroughs in micropaleontology for a long time. Here, we propose a solution for identifying specific taxa of Cambrian microfossils using only a few available specimens by transferring a model pre-trained on natural image datasets to the field of paleontological artificial intelligence. The method employs a 34-layer deep residual neural network as the underlying framework, migrates the ImageNet pre-trained model, freezes the low-layer network parameters and retrains the high-layer parameters to build a microfossil image recognition model. We built training sets with randomly selected images of varied number for each taxon. Our experiments show that the average recognition accuracy for specific taxa of Cambrian microfossils (50 images for each taxon) is higher than 0.97 and it can reach 0.85 with only three training samples per taxon. Comparative analyses indicate that our results are much better than those of various prevalent methods, such as the transpose convolutional neural network (TCNN). This demonstrates the feasibility of using natural images (ImageNet) for the training of microfossil recognition models and provides a promising tool for the discovery of rare fossils.
Keywordearly Cambrian microfossils small sample transfer learning residual network
Indexed BySCI
WOS Research AreaLife Sciences & Biomedicine - Other Topics
WOS SubjectBiology
WOS IDWOS:000914310200001
Document Type期刊论文
Corresponding AuthorYang, Xiaoguang; Zhang, Tao
Affiliation1.Northwest Univ, Sch Informat Sci & Technol, Xian 710069, Peoples R China
2.Northwest Univ, Dept Geol, Shaanxi Key Lab Early Life & Environm, State Key Lab Continental Dynam, Xian 710069, Peoples R China
3.Chinese Acad Sci, Nanjing Inst Geol & Palaeontol, State Key Lab Palaeobiol & Stratig, Nanjing 210008, Peoples R China
Corresponding Author AffilicationNanjing Institute of Geology and Palaeonotology,CAS
Recommended Citation
GB/T 7714
Wang, Bin,Sun, Ruyue,Yang, Xiaoguang,et al. Recognition of Rare Microfossils Using Transfer Learning and Deep Residual Networks[J]. BIOLOGY-BASEL,2023,12(1):14.
APA Wang, Bin.,Sun, Ruyue.,Yang, Xiaoguang.,Niu, Ben.,Zhang, Tao.,...&Han, Jian.(2023).Recognition of Rare Microfossils Using Transfer Learning and Deep Residual Networks.BIOLOGY-BASEL,12(1),14.
MLA Wang, Bin,et al."Recognition of Rare Microfossils Using Transfer Learning and Deep Residual Networks".BIOLOGY-BASEL 12.1(2023):14.
Files in This Item:
File Name/Size DocType Version Access License
biology-12-00016.pdf(3585KB)期刊论文作者接受稿开放获取CC BY-NC-SAApplication Full Text
Related Services
Usage statistics
Google Scholar
Similar articles in Google Scholar
[Wang, Bin]'s Articles
[Sun, Ruyue]'s Articles
[Yang, Xiaoguang]'s Articles
Baidu academic
Similar articles in Baidu academic
[Wang, Bin]'s Articles
[Sun, Ruyue]'s Articles
[Yang, Xiaoguang]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Wang, Bin]'s Articles
[Sun, Ruyue]'s Articles
[Yang, Xiaoguang]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.