Computational modeling demonstrates separation and recognition of overlapping digits using a deep capsule network, highlighting computer vision capabilities in visual segmentation.
Separation of overlapping digits is observed as a very difficult task when done with human eye, but subsequently it becomes easy when computer vision is used for the same task. The authors have constructed a capsule network which recognizes and separates the overlapped digits. Small sized convolutional kernels were used for the first step of feature extraction which provides grained features this reducing the training parameters. The same process could also possibly be applied for overlapping alphabets and characters also. The process was further followed by insertion of capsule layer into the model which resulted in 93.08% accuracy. Finally the overlapped digits were separated and the image of each digit was reconstructed.
No takes yet. Share an insight, caveat, or question.
Moudgil et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: