The Editor-in-Chief has retracted this Article due to concerns about the reliability of the work presented.

Post publication peer review found the following issues:

  • The sections related to depth wish-deformable convolutional neural network structure (DD-CNN) show similarities with Dai et al. (2017) [1], but this source has not been cited;

  • The literature review is incomplete as there is a large volume of underwater image enhancement work which has not been referenced and there is no overview of the state-of-the-art (e.g. YOLO, Faster/Mask R-CNN);

  • There is no detailed explanation of the dataset or examples of input data so the target for this study is not clear and the uneven lighting issues are not shown;

  • A comparison of the methodology with the current state-of-the art is missing;

  • In the Experimental section, for the target recognition performance using the LeNet-5 network modified with deformable convolutional layers placed at different locations, there is no justification for the choice of placement of the deformable layer;

  • There does not seem to be any ablation study being reported and the overall experimental evaluation is weak;

  • In the underwater environment, the performance of the algorithm is different even through the pre-processing of the image, so it is necessary to clarify which part affects the performance enhancement when comparing the results;

  • It is not clear what the tables are showing and the units are missing.

The authors have not responded to correspondence from the publisher about this retraction.

The online version of this article contains the full text of the retracted article as Supplementary Information.


References

[1] Dai, J., Qi, H., Xiong, Y., Li, Y., Zhang, G., Hu, H. and Wei, Y., 2017. Deformable convolutional networks. In Proceedings of the IEEE international conference on computer vision (pp. 764–773)