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

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.

Highly Reliable Method to Obtain a Correlation Coefficient Unaffected by the SEM Noise Component for Examining the Degree of Specimen Damage due to Electron Beam Irradiation

A correlation coefficient is often used as a measure of the strength of a linear relationship (i.e., the degree of similarity) between two sets of data in a variety of fields. However, in the field of scanning electron microscopy (SEM), it is frequently difficult to properly use the correlation coefficient because SEM images generally include severe noise, which affects the measurement of this coefficient. The current study describes a method of obtaining a correlation coefficient that is unaffected by SEM noise in principle. This correlation coefficient is obtained from a total of four SEM images, comprising two sets of two images with identical views, by calculating several covariance values. Numerical experiments confirm that the measured correlation coefficients obtained using the proposed method for noisy images are equal to those for noise-free images. Furthermore, the present method can be combined with a highly accurate and noise-robust position alignment as needed. As one application, we show that it is possible to immediately examine the degree of specimen damage due to electron beam irradiation during a certain SEM observation, which has been difficult until now.