More synthesized results


Figure 1. Synthesized images from models learned on the SVHN dataset. From left to right: observed images, images synthesized by DCGAN (Radford, Metz, and Chintala, 2015), single-grid method and multi-grid method. CD1 and persistent CD cannot synthesize realistic images and their results are not shown.

(a) Observed images

(b) Synthesized images

Figure 2. Synthesized images generated by multi-grid method learned on swimming pool category in MIT places205.

(a) Observed images

(b) Synthesized images

Figure 3. Synthesized images generated by multi-grid method learned on rock category in MIT places205.

(a) Observed images

(b) Synthesized images

Figure 4. Synthesized images generated by multi-grid method learned on building_facade category in MIT places205.

(a) Observed images

(b) Synthesized images

Figure 5. Synthesized images generated by multi-grid method learned on forest road category in MIT places205.

(a) Observed images

(b) Synthesized images

Figure 6. Synthesized images generated by multi-grid method learned on hotel room category in MIT places205.

Acknowledgement

We thank Zhuowen Tu for sharing his insights on his recent work on introspective learning. We thank Jianwen Xie for helpful discussions.

The work is supported by DARPA SIMPLEX N66001-15-C-4035, ONR MURI N00014-16-1-2007, and DARPA ARO W911NF-16-1-0579.

Top
Back to main