ERRATA I n the November 2017 issue of IEEE Signal Processing Magazine, an outdated version of Table 3 was published in the article by Xiaodong He and Li Deng [1]. The updated table is shown below. We sincerely apologize for this error and any confusion it may have caused. Reference [1] X. He and L. Deng, "Deep learning for image-totext generation," IEEE Signal Process. Mag., vol. 34, no. 6, pp. 109-116, Nov. 2017. Table 3. The state-of-the-art image captioning systems in automatic metrics measured using 40 caption references per image (as of 30 August 2017). SPICE CIDEr-D METEOR BLEU-4 Date Human 0.74 0.91 0.335 0.471 23 March 2015 MSR/ACRV 0.715 1.205 0.367 0.685 22 July 2017 DEEPAI 0.711 1.194 0.364 0.67 22 July 2017 TencentVision 0.704 1.224 0.366 0.673 7 August 2017 CASIA_IVA 0.702 1.188 0.362 0.669 22 July 2017 bmc-uestc 0.695 1.046 0.364 0.642 2 August 2017 CAP_BMC 0.693 1.047 0.365 0.645 13 June 2017 SenmaoYe 0.692 1.059 0.37 0.639 29 April 2017 Watson Multimodal 0.689 1.167 0.355 0.645 17 March 2017 DONOT_FAIL_AGAIN 0.683 1.026 0.355 0.612 22 November 2016 QMUL-VISION 0.68 1.121 0.344 0.625 27 June 2017 ucas_yu 0.674 1.028 0.357 0.628 7 June 2017 MetaMind/VT_GT 0.673 1.059 0.359 0.637 1 December 2016 TencentVision 0.67 1.035 0.359 0.64 27 March 2017 MSM@MSRA 0.669 1.053 0.361 0.646 25 October 2016 SRCB@Ricoh 0.664 1.047 0.362 0.651 12 January 2017 LC-JHU 0.657 1.064 0.348 0.617 11 February 2017 NTU_ROSE_CAP 0.657 1.025 0.336 0.586 31 May 2017 UPC2017 0.656 1.02 0.351 0.615 8 May 2017 THU_MIG 0.656 1.013 0.336 0.614 8 June 2017 ATT-IMG/MSRA 0.653 1.036 0.356 0.645 13 June 2016 UTS 0.653 0.972 0.343 0.577 13 June 2017 reviewnet 0.649 0.969 0.347 0.597 24 October 2016 G 0.645 1.012 0.34 0.6 15 March 2017 HUST_2017 0.644 0.989 0.345 0.611 15 March 2017 Digital Object Identifier 10.1109/MSP.2017.2776878 Date of publication: 9 January 2018 178 IEEE SIGNAL PROCESSING MAGAZINE | January 2018 |