DeepFH Segmentations for Superpixel-based Object Proposal Refinement
This work addresses the need for better segmentation accuracy in object proposal generation for computer vision systems, though it is incremental as it builds on existing superpixel and refinement techniques.
The paper tackles the problem of inaccurate object proposal segmentations in detection pipelines by introducing a superpixel-based refinement system that uses DeepFH segmentation, which enhances classic FH segmentation with deep features. On the COCO dataset with LVIS annotations, this approach outperforms state-of-the-art methods, leading to more precise object proposals.
Class-agnostic object proposal generation is an important first step in many object detection pipelines. However, object proposals of modern systems are rather inaccurate in terms of segmentation and only roughly adhere to object boundaries. Since typical refinement steps are usually not applicable to thousands of proposals, we propose a superpixel-based refinement system for object proposal generation systems. Utilizing precise superpixels and superpixel pooling on deep features, we refine initial coarse proposals in an end-to-end learned system. Furthermore, we propose a novel DeepFH segmentation, which enriches the classic Felzenszwalb and Huttenlocher (FH) segmentation with deep features leading to improved segmentation results and better object proposal refinements. On the COCO dataset with LVIS annotations, we show that our refinement based on DeepFH superpixels outperforms state-of-the-art methods and leads to more precise object proposals.