Correcting Imprecise Object Locations for Training Object Detectors in Remote Sensing Applications

Abstract

Object detection on aerial and satellite imagery is an important tool for image analysis in remote sensing and has many areas of application. As modern object detectors require accurate annotations for training, manual and labor-intensive labeling is necessary. In situations where GPS coordinates for the objects of interest are already available, there is potential to avoid the cumbersome annotation process. Unfortunately, GPS coordinates are often not well-aligned with georectified imagery. These spatial errors can be seen as noise regarding the object locations, which may critically harm the training of object detectors and, ultimately, limit their practical applicability. To overcome this issue, we propose a co-correction technique that allows us to robustly train a neural network with noisy object locations and to transform them toward the true locations. When applied as a preprocessing step on noisy annotations, our method greatly improves the performance of existing object detectors. Our method is applicable in scenarios where the images are only annotated with points roughly indicating object locations, instead of entire bounding boxes providing precise information on the object locations and extents. We test our method on three datasets and achieve a substantial improvement (e.g., 29.6% mAP on the COWC dataset) over existing methods for noise-robust object detection.

Improving Object Detectors by Exploiting Bounding Boxes for Augmentation Design

“…This often results in imprecise bounding box labels. To address this challenge of noisy labels, recent works [24]- [27] have developed sophisticated training methods that typically require large amounts of data, computationally expensive optimization strategies and multiple additional objectives. In contrast InterAug works out of the box, without requiring any modifications to the training loop or the model architecture, and provides significantly robust detectors and is effective even under scarce training data scenarios.…”

Weakly-Supervised Salient Object Detection With Saliency Bounding Boxes

“…vision background are not sure how to annotate high-quality boxes, while annotators without domain knowledge can also be difficult to annotate accurate object boxes. For example, recent wheat head detection challenge 1 that was hosted at the European Conference on Computer Vision (ECCV) workshop 2020 has shown that precise object bounding boxes are not easy to obtain, because in some domains the definition of the object is significantly different from generic objects in COCO, thus brings annotation ambiguities (Fig. 1).…”

“…Recently, learning object detectors with noisy data have gained a surge of interest, several approaches [1,5,25,47] have made attempted to tackle noisy annotations. These approaches often assume that the noise occurs both on category labels and bounding box annotations, and devise a disentangled architecture to learn object detectors.…”

“…A major line of research focuses on the classification task, and develops various techniques to deal with noisy labels, such as sample selection [18,21], label correction [32,37], and robust loss functions [16,50]. Recently, much effort [1,5,25,33,47] has been devoted to the object detection task. Simon et al [5] first investigate the impact of different types of label noise on object detection, and propose a per-object co-teaching method to alleviate the effect of noisy labels.…”

Data-centric annotation analysis for plant disease detection: Strategy, consistency, and performance

“…Due to these practical challenges, the actual labels often deviate from the ideal value, which leads to labels that are inconsistent with the instances. Even though extensive works of deep learning techniques under class noise exist ( Li et al., 2020 ; Bernhard and Schubert, 2021 ; Mao et al., 2021 ; Xu et al., 2021 ), it mainly focuses on computer vision datasets such as MS-COCO ( Lin et al., 2014 ), PASCAL VOC ( Everingham et al., 2010 ), and ImageNet ( Deng et al., 2009 ) rather than domain-specific datasets. In some domains, the definition of an object is significantly different from generic objects in these typical datasets, thus bringing annotation ambiguities.…”