Attention Mechanism-Enhanced Model for Automated Simple Brush Stroke Painting

Abstract

With the continuous development of computer technology, the application scope has become increasingly widespread. The flexible application of computer technology related mechanisms in automated simple brush painting technology can be further developed. Therefore, to further optimize the accuracy of automated simple stroke painting technology in generating images, making the images closer to real hand drawn images, a simple stroke painting model based on attention mechanism and long short-term memory network is constructed. The experimental environment is conducted using NVIDIA Tesla K80 GPU, 256GB of memory, and running on Python 3.8.13 and TensorFlow 2. X. On the FaceX dataset, the F1 score and Precision of the AM-LSTM reached 98.75% and 98.63% respectively. Compared to CNN, the F1 score and Precision has increased by 2.01% and 1.40% respectively. The improved attention mechanism model has good performance in automated simple brush painting technology. This indicates that the improved attention machine model can effectively improve image recognition performance. This research method innovatively combines attention mechanisms with long short-term memory network to construct the simple brush painting technology, resulting in more vivid and realistic simple strokes.