Applying of Adaptive Threshold Non-maximum Suppression to Pneumonia Detection

Hyper-parameters in deep learning are sensitive to prediction results. Non-maximum suppression (NMS) is an indispensable method for the object detection pipelines. NMS uses a pre-defined threshold algorithm to suppress the bounding boxes while their overlaps are not significant. We found that the pre-defined threshold is a hyper-parameter determined by empirical knowledge. We propose an adaptive threshold NMS that uses different thresholds to suppress the bounding boxes whose overlaps are not significant. The proposed adaptive threshold NMS algorithm provides improvements on Faster R-CNN with the AP metric on pneumonia dataset. Furthermore, we intend to propose more methods to optimize the hyper-parameters.

[1]  Ali Farhadi,et al.  YOLO9000: Better, Faster, Stronger , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Sebastian Ruder,et al.  An overview of gradient descent optimization algorithms , 2016, Vestnik komp'iuternykh i informatsionnykh tekhnologii.

[3]  Lars Petersson,et al.  Improving Object Localization with Fitness NMS and Bounded IoU Loss , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[4]  Trevor Darrell,et al.  Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation , 2013, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[5]  A. Rosenfeld,et al.  Edge and Curve Detection for Visual Scene Analysis , 1971, IEEE Transactions on Computers.

[6]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[7]  Bernt Schiele,et al.  Learning Non-maximum Suppression , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[9]  Jie Lin,et al.  MaxpoolNMS: Getting Rid of NMS Bottlenecks in Two-Stage Object Detectors , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[10]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  I. Overington,et al.  Practical first-difference edge detection with subpixel accuracy , 1987, Image Vis. Comput..

[12]  Nikhil S. Ketkar.  Stochastic Gradient Descent , 2017 .

[13]  Wei Liu,et al.  SSD: Single Shot MultiBox Detector , 2015, ECCV.

[14]  Kaiming He,et al.  Feature Pyramid Networks for Object Detection , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  D. Spalding.  A novel finite difference formulation for differential expressions involving both first and second derivatives , 1972 .

[16]  Bernt Schiele,et al.  A Convnet for Non-maximum Suppression , 2015, GCPR.

[17]  Yunhong Wang,et al.  Adaptive NMS: Refining Pedestrian Detection in a Crowd , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[18]  Kaiming He,et al.  Focal Loss for Dense Object Detection , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[19]  Hao Chen,et al.  FCOS: Fully Convolutional One-Stage Object Detection , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[20]  Leslie N. Smith,et al.  A disciplined approach to neural network hyper-parameters: Part 1 - learning rate, batch size, momentum, and weight decay , 2018, ArXiv.

[21]  Larry S. Davis,et al.  Soft-NMS — Improving Object Detection with One Line of Code , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[22]  Yi Li,et al.  R-FCN: Object Detection via Region-based Fully Convolutional Networks , 2016, NIPS.

[23]  Ross B. Girshick,et al.  Fast R-CNN , 2015, 1504.08083.

[24]  David G. Lowe,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004, International Journal of Computer Vision.