Skip to main content
. 2021 Oct 19;11:20696. doi: 10.1038/s41598-021-99998-z

Table 4.

Performance comparison of the proposed FER method with some past methods evaluated on IRDatabase.

Work references Method Accuracy (%)
Kopaczka et al.12 HOG-SVM 46.7
Kopaczka et al.40 HOG features with Random Forest classifier (Neutral, Happy, Sad and Surprised) 65.7
Prabhakaran et al.28 ResNet152-V2 79.54
Proposed method IRFacExNet 88.433

Bold value indicates result of the proposed methodology.