Toward Scale-Invariance and Position-Sensitive Region Proposal Networks
This work addresses the need for high-quality object proposals in detection frameworks, offering significant performance gains for computer vision applications, though it is incremental in improving existing region proposal methods.
The paper tackles the problem of accurately localizing object proposals to improve object detection by proposing a scale-invariant and position-sensitive region proposal network, which increases average recall by 35% on PASCAL VOC and 45% on COCO at 1,000 proposals with a fast inference time of 44.8 ms.
Accurately localising object proposals is an important precondition for high detection rate for the state-of-the-art object detection frameworks. The accuracy of an object detection method has been shown highly related to the average recall (AR) of the proposals. In this work, we propose an advanced object proposal network in favour of translation-invariance for objectness classification, translation-variance for bounding box regression, large effective receptive fields for capturing global context and scale-invariance for dealing with a range of object sizes from extremely small to large. The design of the network architecture aims to be simple while being effective and with real time performance. Without bells and whistles the proposed object proposal network significantly improves the AR at 1,000 proposals by $35\%$ and $45\%$ on PASCAL VOC and COCO dataset respectively and has a fast inference time of 44.8 ms for input image size of $640^{2}$. Empirical studies have also shown that the proposed method is class-agnostic to be generalised for general object proposal.