What is Detectron2-FC

Detectron2-FC a fast construction platform of neural network algorithm based on detectron2. We have been working hard in two directions:

1)Quickly build complex deep learning network algorithm for training and prediction;

2)It can quickly build deep learning algorithms for almost all tasks.

Fortunately, detectron2 points out the direction for us. It builds a very convenient algorithm building platform based on pytorch, which can quickly reproduce the latest algorithms. However, detectron2 can only build algorithms with few tasks such as target detection and segmentation. For example, simple image classification cannot find convenience on detectron2. Therefore, on the basis of detectron2, we continue to enrich the functions of detectron2 to make it suitable for image classification, prediction, meta learning and other tasks.

Installation tutorial

See https://detectron2.readthedocs.io/en/latest/tutorials/install.html. The construction environment is as follows:




Existing examples

Image classification


Object detection and image segmentation

Detectron2’s own algorithm see: https://github.com/facebookresearch/detectron2/blob/main/MODEL_ZOO.md

Few-shot object detection


View Github