Fig. 3From: Automatic cardiac evaluations using a deep video object segmentation networkDetails of segmentation and box subnets in the EchoRCNN architecture (inspired from RetinaNet)Back to article page