The main contents of the project include image feature extraction and mesh template
deformation. Key technologies such as VGG16, graph convolution + graph pooling are
adopted, the research results of multi view 3D reconstruction network model are obtained.
The main content and difficulty of the project is to prevent the characters in the
picture from being recognized by intelligent algorithms. Based on faster RCNN and
generated countermeasure network, target detection and privacy protection are
realized respectively, the recognition rate has been improved.
The main content of the project is to reconstruct the corresponding high-resolution
image from the low-resolution image, realize the super-resolution reconstruction
based on the generated countermeasure network, Close to the real picture in effect.
