Simplified FCOS (pytorch)

Re-implementation of FCOS: Fully Convolutional One-Stage Object Detection (https://arxiv.org/abs/1904.01355)

---

Modified codes from:

https://github.com/rosinality/fcos-pytorch

What's different?

• Simplified as I can to increase legibility
• Implementing in-house (tooth detection) data loader instead of COCO dataset.
• Removing parallel computing functions

---

Backbone

Using Vovnet57 as backbone Pre-trained pth download from here(ref)

---

TO-DO:

• Add modified evalute funtion for in-house data
• Add image output function for visualizations