IQDet: Instance-wise Quality Distribution Sampling for Object Detection
This work addresses the challenge of selecting high-quality training samples in object detection, offering a cost-free improvement for applications like autonomous driving and surveillance.
The paper tackles the problem of dense object detection by proposing an instance-wise quality distribution sampling strategy, which improves baseline performance by 2.4 AP on MS COCO and achieves 51.6 AP, outperforming existing one-stage detectors without increasing inference cost.
We propose a dense object detector with an instance-wise sampling strategy, named IQDet. Instead of using human prior sampling strategies, we first extract the regional feature of each ground-truth to estimate the instance-wise quality distribution. According to a mixture model in spatial dimensions, the distribution is more noise-robust and adapted to the semantic pattern of each instance. Based on the distribution, we propose a quality sampling strategy, which automatically selects training samples in a probabilistic manner and trains with more high-quality samples. Extensive experiments on MS COCO show that our method steadily improves baseline by nearly 2.4 AP without bells and whistles. Moreover, our best model achieves 51.6 AP, outperforming all existing state-of-the-art one-stage detectors and it is completely cost-free in inference time.