- Contains scripts to
- Utils.py: Convert Dataset from Masks to COCO fromate (CMD Interface Not Avialable)
- Visualizer.py: Creates Visualisation for generated COCO formate (CMD Interface Not Avialable)
- Dataset_divide.py: Divides files into given number and saves in required dataset (CMD Avialable)
- Helper.py: Script contains functions for Random Seeding for Pytroch ENV, Creating Directory if does not exist, epoch_time- to calculate the time taken for epochs.
For traning segmentation model, Pretrained model from Detectron2 Pytorch framwork was been imported.
Mask RCNN along with FPN with RESNET 50 as backbone archetecture was been used.
- Detectron2_Base: Notebook for traning model. Saves visualisation in /plots directory
- model_2: Contains model file, and segmentation all segmentation outputs and results.
