Understanding cartoon emotion using integrated deep neural network on large dataset.

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Emotion recognition is crucial for understanding human-computer interaction.
• Existing research primarily focuses on real human faces, leaving cartoon-based emotion analysis underdeveloped.
• Analyzing emotions in cartoons presents unique challenges due to stylized representations.

Purpose of the Study:

• To develop an integrated Deep Neural Network (DNN) approach for accurate emotion recognition in cartoon images.
• To address the data scarcity in cartoon emotion recognition by creating a novel dataset.
• To evaluate the performance of various deep learning models for emotion classification in cartoons.

Main Methods:

• A custom dataset of 8,000 images featuring 'Tom' and 'Jerry' with four emotions (happy, sad, angry, surprise) was curated.
• Mask R-CNN was employed for character detection and face mask segmentation.
• Deep learning models including ResNet-50, MobileNetV2, InceptionV3, and VGG 16 were utilized for emotion classification.

Main Results:

• The integrated DNN approach achieved an overall accuracy of 0.96 in recognizing emotions from cartoon characters.
• VGG 16 demonstrated superior performance among the tested models, yielding an accuracy of 96% and an F1 score of 0.85 for emotion classification.
• The proposed method outperformed existing state-of-the-art approaches in cartoon emotion recognition.

Conclusions:

• The developed integrated DNN approach is effective for emotion recognition in cartoon images.
• The VGG 16 model shows significant promise for classifying emotions in stylized visual data.
• This research contributes a valuable dataset and a robust methodology for advancing cartoon emotion analysis.