- Adam Szummer - a.szummer@gmail.com - VanillaTiger
- Sergi Garcia Sarroca - sergi.garciasa@e-campus.uab.cat - SoftonCV
- Antoni Rodriguez Villegas - rv.antoni@hotmail.com - antoniRodriguez
Week 5: Hands on Detectron -- Object Detection and Instance Segmentation with Out of Context Datasets.
Main tasks:
(a) Out-of-Context Dataset:
- Get qualitative results on Out-of-Context Dataset
- Analyze the results. Why do you think it fails if it fails?
(b) MSCOCO Dataset by transplanting new objects by co-occurrence:
- Transplant objects from test set of COCO
- Transplanting New Object according to Co-occurrence of objects. Refer to Figure 1 in original paper
(c) MSCOCO Dataset by qualitatively Transplant:
- Transplant objects from test set of COCO
- Qualitatively Transplant Existing Object - Refer to Figure 2 in original paper
(d) MSCOCO Dataset by Feature interference:
-
Transplant objects from test set of COCO
-
Feature interference according to Co-occurrence of objects - Refer to Figure 3 in original paper
-
Slides for the project: T00-Google Slides
-
Link to the Overleaf article (non-editable): Group00-Overleaf