FCN (Fully Convolutional Network) is deep fully convolutional neural network architecture for semantic pixel-wise segmentation. This is implementation of "https://arxiv.org/abs/1605.06211" by using Chainer which is a neural networks library. FCN can train by using any size of image, but I trained this network using the images of the same size (256 * 256).
You can download the parameter from here.
!Caution: I trained the model only 100 epochs. If you wanna use this model practicaly, you have to do fine-tuning.
$ python train.py -g <gpu:id> -tr <path to train dataset> -ta <path to target dataset> -tt <image file names text> -e <epoch> -b <batchsize>
You can select else options, please read script directly.
$ pyton predict.py -g <gpu:id> -i <path to image> -w <path to weight>
- person
- horse
- bus
