MISC: Multi-Condition Injection and Spatially-Adaptive Compositing for Conditional Person Image Synthesis
Shuchen Weng, Wenbo Li, Dawei Li, Hongxia Jin, Boxin Shi
In this paper, we explore synthesizing person images with multiple conditions for various backgrounds. To this end, we propose a framework named "MISC" for conditional image generation and image compositing. For conditional image generation, we improve the existing condition injection mechanisms by leveraging the inter-condition correlations. For the image compositing, we theoretically prove the weaknesses of the cutting-edge methods, and make it more robust by removing the spatially-invariance constraint, and enabling the bounding mechanism and the spatial adaptability. We show the effectiveness of our method on the Video Instance-level Parsing dataset, and demonstrate the robustness through controllability tests.
