TY  - JOUR
AU  - Sarvakar, Ketan
AU  - Rana, Kaushik
PY  - 2025
DA  - 2025/09/26
TI  - A Hybrid Framework Combining CNN, LSTM, and Transfer Learning for Emotion Recognition
JO  - ICCK Transactions on Machine Intelligence
T2  - ICCK Transactions on Machine Intelligence
JF  - ICCK Transactions on Machine Intelligence
VL  - 1
IS  - 2
SP  - 103
EP  - 116
DO  - 10.62762/TMI.2025.572412
UR  - https://www.icck.org/article/abs/TMI.2025.572412
KW  - face expressions
KW  - face emotion recognition
KW  - deep learning
KW  - CNN
KW  - VGG-16
KW  - ResNet-50
KW  - LSTM
KW  - transfer learning
AB  - Deep learning has substantially enhanced facial emotion recognition, an essential element of human--computer interaction. This study evaluates the performance of multiple architectures, including a custom CNN, VGG-16, ResNet-50, and a hybrid CNN-LSTM framework, across FER2013 and CK+ datasets. Preprocessing steps involved grayscale conversion, image resizing, and pixel normalization. Experimental results show that ResNet-50 achieved the highest accuracy on FER2013 (76.85%), while the hybrid CNN-LSTM model attained superior performance on CK+ (92.30%). Performance metrics such as precision, recall, and F1-score were used for evaluation. Findings highlight the trade-off between computational efficiency and recognition accuracy, offering insights for developing robust, real-time emotion recognition systems.
SN  - 3068-7403
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Sarvakar2025A,
  author = {Ketan Sarvakar and Kaushik Rana},
  title = {A Hybrid Framework Combining CNN, LSTM, and Transfer Learning for Emotion Recognition},
  journal = {ICCK Transactions on Machine Intelligence},
  year = {2025},
  volume = {1},
  number = {2},
  pages = {103-116},
  doi = {10.62762/TMI.2025.572412},
  url = {https://www.icck.org/article/abs/TMI.2025.572412},
  abstract = {Deep learning has substantially enhanced facial emotion recognition, an essential element of human--computer interaction. This study evaluates the performance of multiple architectures, including a custom CNN, VGG-16, ResNet-50, and a hybrid CNN-LSTM framework, across FER2013 and CK+ datasets. Preprocessing steps involved grayscale conversion, image resizing, and pixel normalization. Experimental results show that ResNet-50 achieved the highest accuracy on FER2013 (76.85\%), while the hybrid CNN-LSTM model attained superior performance on CK+ (92.30\%). Performance metrics such as precision, recall, and F1-score were used for evaluation. Findings highlight the trade-off between computational efficiency and recognition accuracy, offering insights for developing robust, real-time emotion recognition systems.},
  keywords = {face expressions, face emotion recognition, deep learning, CNN, VGG-16, ResNet-50, LSTM, transfer learning},
  issn = {3068-7403},
  publisher = {Institute of Central Computation and Knowledge}
}