Ground-truth Boxes Research Articles

Hot Anchors: A Heuristic Anchors Sampling Method in RCNN-Based Object Detection.

In the image object detection task, a huge number of candidate boxes are generated to match with a relatively very small amount of ground-truth boxes, and through this method the learning samples can be created. But in fact the vast majority of the candidate boxes do not contain valid object instances and should be recognized and rejected during the training and evaluation of the network. This leads to extra high computation burden and a serious imbalance problem between object and none-object samples, thereby impeding the algorithm’s performance. Here we propose a new heuristic sampling method to generate candidate boxes for two-stage detection algorithms. It is generally applicable to the current two-stage detection algorithms to improve their detection performance. Experiments on COCO dataset showed that, relative to the baseline model, this new method could significantly increase the detection accuracy and efficiency.

Hierarchical objectness network for region proposal generation and object detection

Recent region proposal generation methods show a low Intersection-of-Union with the ground-truth boxes. Because they simply regress the coordinates of the bounding boxes by exploiting the single-layer output of convolutional neural networks. This paper proposes a hierarchical objectness network for region proposal generation and object detection to address the inaccurate localization problem. Instead of regressing the coordinates, we subtly localize the objects by predicting the stripe objectness, i.e., a group of probabilities reflecting the existence of the object in each location of the candidate proposal. Additionally, we construct the hierarchical features by reversely connecting multiple convolutional layers to detect objects with large-scale variations. Our experimental results demonstrate that our method performs better than the state-of-the-art region proposal generation methods in terms of recall. Moreover, by integrating with advanced object detection frameworks, our method achieves superior object detection results.

High resolution acoustic characterization and change detection to evaluate hurricane effects on a coastal bay bottom

Using high-resolution acoustics (sidescan sonar, chirp sonar subbottom, and echo sounder profile data) the shallow surface and subsurface of Sister Lake (Terrebonne Basin) of coastal Louisiana’s deltaic plain was imaged pre- and post-hurricanes Katrina and Rita. The pre-hurricanes survey in May 2005 imaged the entire lake bottom. To assess the effects of hurricanes Katrina and Rita, a second survey in October 2005 included acoustic resurveying of corridors of data covering all major sectors of the lake. Regions of sidescan reflectance differences between pre- and post-storm periods, corroborated with regions of onlapping sediment or zones of scour in the chirp sonar subbottom data, show areas both of sedimentation and erosion between surveys. Ground truth box cores were collected to calibrate acoustic data. Preliminary image processing results of the georeferenced pre- and post-storms sidescan datasets indicate 60% of the area experienced sedimentation and 40% was scoured between surveys. Amplitude reflectivity of chirp horizons were mapped, and incorporated with sidescan amplitude and texture attributes in change detection analysis. Standardized digital processing, supervised classification, and change detection techniques are developed to characterize the bay bottom and shallow subsurface and to assess and quantify storm effects. [Work supported by NOAA NA03NMF4520316.]