Hypercorrelation Squeeze for Few-Shot Segmentation

4 Apr 2021  ·  Juhong Min, Dahyun Kang, Minsu Cho ·

Few-shot semantic segmentation aims at learning to segment a target object from a query image using only a few annotated support images of the target class. This challenging task requires to understand diverse levels of visual cues and analyze fine-grained correspondence relations between the query and the support images. To address the problem, we propose Hypercorrelation Squeeze Networks (HSNet) that leverages multi-level feature correlation and efficient 4D convolutions. It extracts diverse features from different levels of intermediate convolutional layers and constructs a collection of 4D correlation tensors, i.e., hypercorrelations. Using efficient center-pivot 4D convolutions in a pyramidal architecture, the method gradually squeezes high-level semantic and low-level geometric cues of the hypercorrelation into precise segmentation masks in coarse-to-fine manner. The significant performance improvements on standard few-shot segmentation benchmarks of PASCAL-5i, COCO-20i, and FSS-1000 verify the efficacy of the proposed method.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Benchmark
Few-Shot Semantic Segmentation COCO-20i (10-shot) HSNet Mean IoU 48.7 # 1
Few-Shot Semantic Segmentation COCO-20i (1-shot) HSNet Mean IoU 41.2 # 2
Few-Shot Semantic Segmentation COCO-20i (5-shot) HSNet Mean IoU 49.5 # 1
Few-Shot Semantic Segmentation FSS-1000 HSNet Mean IoU 86.5 # 3
Few-Shot Semantic Segmentation FSS-1000 (1-shot) HSNet Mean IoU 86.5 # 2
Few-Shot Semantic Segmentation FSS-1000 (5-shot) HSNet Mean IoU 88.5 # 2
Few-Shot Semantic Segmentation PASCAL-5i (10-Shot) HSNet Mean IoU 70.6 # 1
Few-Shot Semantic Segmentation PASCAL-5i (1-Shot) HSNet Mean IoU 66.2 # 2
Few-Shot Semantic Segmentation PASCAL-5i (5-Shot) HSNet Mean IoU 70.4 # 2

Methods