TY - GEN
T1 - Using Bag-of-Visual Words to Classify Intestinal Parasites in Reptiles from Stool Images
AU - Yanascual, Guillermo
AU - Parra, Carla
AU - Grijalva, Felipe
AU - Benítez, Diego
AU - Pérez, Noel
AU - Párraga-Villamar, Viviana
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Captive reptile environments promote parasite proliferation, leading to infections and weak immune responses. Furthermore, the manual analysis of stool samples for parasite detection is laborious and susceptible to errors. Therefore, this paper proposes a Bag-of-Visual Word approach to classify six captive reptile parasitic agents, namely Ophionyssus natricis, Blastocystis sp, Oxiurdo egg, Rhytidoides similis, Strongyloides and Taenia, and identify the absence of such parasites from microscope stool images. Based on relevant key points and the SURF descriptor, the Bag-of-Visual Word representation is integrated with a support vector machine-based classifier. We trained the model on a publicly available dataset composed of 3616 images labeled by experts. Experimental results show competitive performance (84.4% accuracy) compared to computationally intensive methods such as conventional convolutional neural networks.
AB - Captive reptile environments promote parasite proliferation, leading to infections and weak immune responses. Furthermore, the manual analysis of stool samples for parasite detection is laborious and susceptible to errors. Therefore, this paper proposes a Bag-of-Visual Word approach to classify six captive reptile parasitic agents, namely Ophionyssus natricis, Blastocystis sp, Oxiurdo egg, Rhytidoides similis, Strongyloides and Taenia, and identify the absence of such parasites from microscope stool images. Based on relevant key points and the SURF descriptor, the Bag-of-Visual Word representation is integrated with a support vector machine-based classifier. We trained the model on a publicly available dataset composed of 3616 images labeled by experts. Experimental results show competitive performance (84.4% accuracy) compared to computationally intensive methods such as conventional convolutional neural networks.
KW - Bag-of-visual-words (BoVW)
KW - Parasites classification
KW - SVM
KW - reptiles
UR - https://www.scopus.com/pages/publications/85179507985
U2 - 10.1109/ETCM58927.2023.10309023
DO - 10.1109/ETCM58927.2023.10309023
M3 - Contribución a la conferencia
AN - SCOPUS:85179507985
T3 - ECTM 2023 - 2023 IEEE 7th Ecuador Technical Chapters Meeting
BT - ECTM 2023 - 2023 IEEE 7th Ecuador Technical Chapters Meeting
A2 - Lalaleo, David Rivas
A2 - Chauvin, Manuel Ignacio Ayala
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 7th IEEE Ecuador Technical Chapters Meeting, ECTM 2023
Y2 - 10 October 2023 through 13 October 2023
ER -