Merging CK+, FER+, and KDEF datasets with online/offline augmentation and random weighted sampling enables a deep CNN to classify seven facial emotions at 82% accuracy.
Matsumoto, More evidence for the universality of a contempt ex- pression, Motivation and Emotion 16 (1992) 363–368
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Improving Facial Emotion Recognition through Dataset Merging and Balanced Training Strategies
Merging CK+, FER+, and KDEF datasets with online/offline augmentation and random weighted sampling enables a deep CNN to classify seven facial emotions at 82% accuracy.