An implementation of Single-Shot-Detector that uses circular anchors for training and outputs circular bounding boxes. The summary slides are under the root folder in both the PPTx and PDF formats. The code is under the code/ directory. The model is trained on the voc_2012_trainval and validated on voc_2007_test. After training, run demo.py to see inference results.
Code foundation credit to: @lufficc https://github.com/lufficc/SSD