CRAFT Objects from Images

TL;DR

CRAFT introduces a cascade approach to object detection that improves proposal quality and classification accuracy by further dividing the tasks into sub-tasks, leading to state-of-the-art results.

Contribution

It proposes a novel cascade network architecture that enhances object proposal generation and classification, advancing the state-of-the-art in object detection.

Findings

Improved object proposal localization and compactness.

Reduced false positives in classification.

Achieved superior performance on PASCAL VOC and ILSVRC benchmarks.

Abstract

Object detection is a fundamental problem in image understanding. One popular solution is the R-CNN framework and its fast versions. They decompose the object detection problem into two cascaded easier tasks: 1) generating object proposals from images, 2) classifying proposals into various object categories. Despite that we are handling with two relatively easier tasks, they are not solved perfectly and there's still room for improvement. In this paper, we push the "divide and conquer" solution even further by dividing each task into two sub-tasks. We call the proposed method "CRAFT" (Cascade Region-proposal-network And FasT-rcnn), which tackles each task with a carefully designed network cascade. We show that the cascade structure helps in both tasks: in proposal generation, it provides more compact and better localized object proposals; in object classification, it reduces false positives (mainly between ambiguous categories) by capturing both inter- and intra-category variances. CRAFT achieves consistent and considerable improvement over the state-of-the-art on object detection benchmarks like PASCAL VOC 07/12 and ILSVRC.