In this technical report, we present key details of our winning panoptic\nsegmentation architecture EffPS_b1bs4_RVC. Our network is a lightweight version\nof our state-of-the-art EfficientPS architecture that consists of our proposed\nshared backbone with a modified EfficientNet-B5 model as the encoder, followed\nby the 2-way FPN to learn semantically rich multi-scale features. It consists\nof two task-specific heads, a modified Mask R-CNN instance head and our novel\nsemantic segmentation head that processes features of different scales with\nspecialized modules for coherent feature refinement. Finally, our proposed\npanoptic fusion module adaptively fuses logits from each of the heads to yield\nthe panoptic segmentation output. The Robust Vision Challenge 2020 benchmarking\nresults show that our model is ranked #1 on Microsoft COCO, VIPER and WildDash,\nand is ranked #2 on Cityscapes and Mapillary Vistas, thereby achieving the\noverall rank #1 for the panoptic segmentation task.\n