MSE-CNN Implementation 1
Code database with the implementation of MSE-CNN, from the paper 'DeepQTMT: A Deep Learning Approach for Fast QTMT-based CU Partition of Intra-mode VVC'
Loading...
Searching...
No Matches
Namespaces | Functions | Variables
demo.py File Reference

Namespaces

namespace  demo
 

Functions

 demo.setup_model ()
 Initializes and load the parameters of the MSE-CNN.
 
 demo.int2label (split)
 Obtain the string that corresponds to an integer value of the split.
 
 demo.draw_partition (img, split, cu_pos, cu_size)
 Draw partition in image based in the split outputed by the model.
 
 demo.split_fm (cu, cu_pos, split)
 Splits feature maps in specific way.
 
 demo.partition_img (img, img_yuv)
 Partitions a full 128x128 CTU and draws the partition in the original image.
 
 demo.pipeline (img, text)
 Pipeline to implement the functionalities to demonstrate the potential of the MSE-CNN.
 
 demo.main ()
 

Variables

str demo.PATH_TO_COEFFS = "../../../model_coefficients/best_coefficients"
 
str demo.LOAD_IMAGE_ERROR = "load_image_error.png"
 
list demo.EXAMPLE_IMGS = ["example_img_1.jpeg", "example_img_2.jpeg"]
 
tuple demo.CTU_SIZE = (128, 128)
 
 demo.FIRST_CU_POS = torch.tensor([0, 0]).reshape(shape=(-1, 2))
 
 demo.FIRST_CU_SIZE = torch.tensor([64, 64]).reshape(shape=(-1, 2))
 
str demo.DEV = "cuda" if torch.cuda.is_available() else "cpu"
 
int demo.QP = 32
 
 demo.model = None
 
tuple demo.COLOR = (0, 247, 255)
 
int demo.LINE_THICKNESS = 1
 
str demo.DEFAULT_TEXT_FOR_COORDS = "Insert CTU position in the image..."