RFAConv: Receptive-field attention convolution for improving convolutional neural networks

Key Points

• Enhancing convolutional neural networks boosts feature extraction, leading to improved performance in image recognition tasks.
• Performance increased by 12% in accuracy on benchmark datasets, demonstrating the method's effectiveness.
• Observational analysis across various convolutional neural networks shows the effectiveness of receptive-field attention techniques.
• Highlighting the need for further exploration of attention mechanisms in neural networks for better applications.