Skip to content

Latest commit

 

History

History
30 lines (24 loc) · 1.67 KB

readme.md

File metadata and controls

30 lines (24 loc) · 1.67 KB
Raw

FCN: Fully Convolutional Networks for Pigs Segmentation

Responsible for segmenting the pigs out of the pictures, that is, removing the background and leaving only the pigs individually.

The structure diagram of FCN.

The structure diagram of FCN

The scalar of Tensorboard of the training process of FCN.

The scalar of Tensorboard of the training process of FCN

The image of Tensorboard of the training process of FCN.

Among them, the "input_image" is the input image, the "output_image" is the image of pigs after segmentation and the "truth mask" is the image of ground truth.

The image 1 of Tensorboard of the training process of FCN The image 2 of Tensorboard of the training process of FCN The image 3 of Tensorboard of the training process of FCN

The screenshot of the CMD of the training process of FCN.

The screenshot of the CMD of the training process of FCN

File Structure.

  • load.py: The base class for loading data.
  • fcn.py: Fcn model. It inherits from "lib/base". You can train the model by running the function FCN.run.
  • get_image.py: It applies fcn.py to segment the pigs of data/TrainImg individually.
  • get_test_image.py: It applies fcn.py to segment the pigs of data/TestB individually.