Thanks @tkianai for your implementation Detectron2 as classification model with MobileNet backbone. Assumpt that we finished training MobileNet backbone for classification task, can we use the trained backbone (with it's weight) as a backbone for Faster R-CNN in Detectron2 ? If yes, what modification in config we need to consider ? Thanks.
Thanks @tkianai for your implementation Detectron2 as classification model with MobileNet backbone. Assumpt that we finished training MobileNet backbone for classification task, can we use the trained backbone (with it's weight) as a backbone for Faster R-CNN in Detectron2 ? If yes, what modification in config we need to consider ? Thanks.