Skin legion segmentation using Deeplab V3

Implementation of DeepLab V3 using fine tuning in pytorch.

Deeplab v3 Architecture

EPOCH: 1/15

Train loss: 0.706819, Test loss: 0.6546

after saving tensor(0.6546, device='cuda:0')

EPOCH: 2/15

Train loss: 0.655163, Test loss: 0.6450

EPOCH: 3/15

Train loss: 0.644734, Test loss: 0.6419

EPOCH: 4/15

Train loss: 0.638355, Test loss: 0.6406

EPOCH: 5/15

Train loss: 0.635301, Test loss: 0.6412

EPOCH: 6/15

Train loss: 0.632920, Test loss: 0.6409

EPOCH: 7/15

Train loss: 0.632391, Test loss: 0.6430

EPOCH: 8/15

Train loss: 0.630760, Test loss: 0.6393

EPOCH: 9/15

Train loss: 0.629521, Test loss: 0.6389

EPOCH: 10/15

Train loss: 0.628565, Test loss: 0.6395

EPOCH: 11/15

Train loss: 0.628196, Test loss: 0.6409

EPOCH: 12/15

Train loss: 0.628002, Test loss: 0.6413

EPOCH: 13/15

Train loss: 0.628979, Test loss: 0.6418

EPOCH: 14/15

Train loss: 0.627591, Test loss: 0.6414

EPOCH: 15/15

Train loss: 0.626776, Test loss: 0.6401