Effective Facial Expression Recognition System Using Machine Learning
Dheeraj Hebri,Ramesh Nuthakki,Ashok Kumar Digal,K G S Venkatesan,Sonam Chawla,C Raghavendra Reddy
DOI: https://doi.org/10.4108/eetiot.5362
2024-03-12
EAI Endorsed Transactions on Internet of Things
Abstract:The co Facial expression recognition (FER) is a topic that has seen a lot of study in computer vision and machine learning. In recent years, deep learning techniques have shown remarkable progress on FER tasks. With this abstract, A Novel Is Advised By Us FER method that combines combined use of k-nearest neighbours and long short-term memory algorithms better efficiency and accurate facial expression recognition. The proposed system features two primary steps—feature extraction and classification—to get results. When extracting features, we extract features from the facial images using the Local Binary Patterns (LBP) algorithm. LBP is a simple yet powerful feature extraction technique that captures texture information from the image. In the classification stage, we use the KNN and LSTM algorithms for facial expression recognition. KNN is a simple and effective classification algorithm that finds the k closest to the given value neighbours to the test training-set-sample and assigning it to the class that is most frequent among its neighbours. However, KNN has limitations in handling temporal information. To address this limitation, we propose to use LSTM, which is a subclass of RNNs capable of capturing temporal relationships in time series data. The LSTM network takes as input the LBP features of a sequence of facial images and processes them through a series of LSTM cells to estimate the ultimate coding of the phrase. We examine the planned and system on two publicly available records: the CK+ and the Oulu-CASIA datasets. According on the experimental findings, the proposed system achieves performance at the cutting edge on both datasets. The proposed system performs better than other state-of-the-art methods, including those that use deep learning systems, quantitatively, in terms of F1-score and precision.In conclusion, the proposed FER system that combines KNN and LSTM algorithms achieves high accuracy and an F1 score in recognising facial expressions from sequences of images. This system can be used in many contexts, including human-computer interaction, emotion detection, and behaviour analysis.