January 2020
tl;dr: Add recurrent LSTM to retinaNet
This paper is quite similar to recurrent SSD but much less insightful. They added two layers of recurrent LSTM to the feature map before the detection head.
K=5 frames
January 2020
tl;dr: Add recurrent LSTM to retinaNet
This paper is quite similar to recurrent SSD but much less insightful. They added two layers of recurrent LSTM to the feature map before the detection head.
K=5 frames