A juvenile ornithopod tracksite from the Lower Cretaceous Haman Formation, South Korea

Yoon, Han Sang; Lee, Yuong‐Nam; Jung, Seung-Ho; Kong, Dehui; Kim, Su Hwan; Son, Minyoung

doi:10.1016/j.cretres.2021.104877

Cited by 3 publications

(5 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, when we enhanced the input image, as shown in Fig 5B , the model classified the footprint as Iguanodontipus with a probability of 99.48% and the lowest probability of 0.03% for Caririchnium . This discrepancy between the literature [ 26 ] and the prediction might result from the input image showing an elongated and narrow digit impression of Iguanodontipus , rather than a short, wide digit impression of Caririchnium . This result implies the machine learning model can be used as an objective means to validate the classification.…”

Section: Resultscontrasting

confidence: 66%

“…For example, although the Hadrosauropodus footprints reported in Tetori, Japan [ 55 ] were not included in the training data, the model classified the given data with a high probability to this ichnotaxon. However, the Caririchnium footprint reported in Gunbuk, Republic of Korea [ 26 ] showed a substantially different classification result (Iguanodontipus) between the model and literature. As the Gunbuk footprint has an elongated and narrow digit impression ( Fig 5B ), the classification results appear to be consistent with the standards for large ornithopod ichnotaxa [ 3 ].…”

Section: Resultsmentioning

confidence: 91%

“… A, Caririchnium from Uiesung, Republic of Korea [ 16 ]; B, High quality illustration of Caririchnium from Gunbuk, Republic of Korea [ 26 ]; C, Low quality illustration of Caririchnium from Gunbuk, Republic of Korea [ 26 ]; D, Hadrosauropodus from La Llau de la Costa, Spain [ 51 ]; E, Hadrosauropodus from Tetori, Japan [ 55 ]; F, Therangospodus oncalensis from Tierras Amigas, Spain [ 13 ]; G, Ornithopodichnus from Sichuan, China [ 58 ]; H, Iguanodontipus track from Soria, Spain [ 52 ] (the footprint is illustrated with multiple lines); I, Ornithopod footprint from Qijiang, China [ 10 ]. …”

Section: Resultsmentioning

confidence: 99%

“…The dinosaur footprint tracks at Gunbuk-myeon, Gyeongsangbuk-do, South Korea, have been reported to be from Caririchnium [ 26 ]. In this example, the original image from the given literature had a low-quality footprint image.…”

Section: Resultsmentioning

confidence: 99%

“…To circumvent such issues, 3D techniques such as photogrammetry and light detection and ranging (LiDAR) have recently been introduced for digitizing and recording dinosaur footprints [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26]. Furthermore, Lallensack [27] produced a program to create objective track outlines from 3D footprints in R environment.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Classification of large ornithopod dinosaur footprints using Xception transfer learning

Ha,

Kim

2023

PLoS ONE

View full text Add to dashboard Cite

Large ornithopod dinosaur footprints have been confirmed on all continents except Antarctica since the 19th century. However, oversplitting problems in ichnotaxa have historically been observed in these footprints. To address these issues and distinguish between validated ichnotaxa, this study employed convolutional neural network-based Xception transfer learning to automatically classify ornithopod dinosaur tracks. The machine learning model was trained for 162 epochs (i.e., the number of full cycles of all training data through the model) using 274 data images, excluding horizontally flipped images. The trained model accuracy was 96.36%, and the validation accuracy was 92.59%. We demonstrate the performance of the machine learning model using footprint illustrations that are not included in the training dataset. These results show that the machine learning model developed in this study can properly classify footprint illustration data for large ornithopod dinosaurs. However, the quality of footprint illustration data (or images) inherently affects the performance of our machine learning model, which performs better on well-preserved footprints. In addition, because the developed machine-learning model is a typical supervised learning model, it is not possible to introduce a new label or class. Although this study used illustrations rather than photos or 3D data, it is the first application of machine-learning techniques at the academic level for verifying the ichnotaxonic assignments of large ornithopod dinosaur footprints. Furthermore, the machine learning model will likely aid researchers to classify the large ornithopod dinosaur footprint ichnotaxa, thereby safeguarding against the oversplitting problem.

show abstract

Section: Resultscontrasting

confidence: 66%