Semantic Pixel Distances for Image Editing

Myers-Dean, Josh; Wehrwein, Scott

doi:10.1109/cvprw50498.2020.00275

Cited by 3 publications

(1 citation statement)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The fundamental computer vision task of semantic segmentation seeks to assign class labels to specific pixels in a given input image. The rapid development of deep learning methods has resulted in significant progress in performance [85,17,83,49], stability [61,79], and accessibility [57,1] of semantic segmentation algorithms, often seen in real world applications such as autonomous vehicles [88,37,72,12,10,53], precision agriculture [5,4], medical diagnosis [82,73,39,60,67], image restoration and editing [54,44], sports [18], and remote sensing [6,41]. While such algorithms provide insightful information about the scene, it requires large amounts of pixel-wise labeled data [87,48,23], which is often expensive and time consuming to collect [9].…”

Section: Introductionmentioning

confidence: 99%

Towards Single Stage Weakly Supervised Semantic Segmentation

Akiva¹,

Dana²

2021

Preprint

View full text Add to dashboard Cite

The costly process of obtaining semantic segmentation labels has driven research towards weakly supervised semantic segmentation (WSSS) methods, using only imagelevel, point, or box labels. The lack of dense scene representation requires methods to increase complexity to obtain additional semantic information about the scene, often done through multiple stages of training and refinement. Current state-of-the-art (SOTA) models leverage image-level labels to produce class activation maps (CAMs) which go through multiple stages of refinement before they are thresholded to make pseudo-masks for supervision. The multi-stage approach is computationally expensive, and dependency on image-level labels for CAMs generation lacks generalizability to more complex scenes. In contrary, our method offers a single-stage approach generalizable to arbitrary dataset, that is trainable from scratch, without any dependency on pre-trained backbones, classification, or separate refinement tasks. We utilize point annotations to generate reliable, on-the-fly pseudo-masks through refined and filtered features. While our method requires point annotations that are only slightly more expensive than image-level annotations, we are to demonstrate SOTA performance on benchmark datasets (PascalVOC 2012), as well as significantly outperform other SOTA WSSS methods on recent real-world datasets (CRAID, CityPersons, IAD).

show abstract