ranju2015 / scene-parsing-framework-with-context-adaptive-feature-intregation Goto Github PK
View Code? Open in Web Editor NEWThe proposed model can be divided into 3 logical layers or modules. The first layer (i.e., the Visual Layer) is introduced to extract optimized features using a Genetic Algorithm and train the One-vs-All binary classifier. The following layer deals with contextual information that learns the global and local contexts of an image. The final layer combines all the information optimally using a noble MLP-based regression method to produce the final class label.