Research Project

## 1. Deep extreme cut: From extreme points to object segmentation, Kevis et al_ CVPR2018 ##    (1) paper and code     paper link: https://arxiv.org/abs/1711.09081     code(pytorch):  https://github.com/scaelles/DEXTR-PyTorch     (website) [ http://people.ee.ethz.ch/~cvlsegmentation/dextr/ ]    (2) problem range     target:  single image testing datasets: COCO, Pascal VOC, GrabCut, Davis 2016, Davis 2017 (3) architecture     (inference) input: four extreme points(left-most, right-most, top and bottom)     (refinement)input: the above four points+   one extra point    (4)Implement details:    Balance loss(cross-entropy)    (5) results:    1> Obtaining state-of-the-art results in all scenarios.    2> Reducing labeling costs by a factor of 10. ##================================================== ## ## 2. Annot...

Matlab2018: /usr/local/MATLAB docker path: /home/uni/Lab/Tools

1. introduction to basic operations https://guides.github.com/activities/hello-world/ 2. uploading an existing project to github https://help.github.com/articles/adding-an-existing-project-to-github-using-the-command-line/ 3. Learn Markdown https://bitbucket.org/tutorials/markdowndemo

##-------------------------------------------------single frame----------------------------------------------------# 1. Path Aggregation Network for Instance Segmentation 1.PANet: Path aggregation network for instance segmentation_Shu Liu_CVPR 2018 paper link:  https://arxiv.org/abs/1803.01534 source code: code will be available ( caffe based) 1> Framework: 2> Contributions: 1bottom-up network structure (1) Bottom-up Path Augmentation: shorten the distance among lower and topmost feature levels for reliable information passing (2) Adaptive Feature Pooling:  pool features from all feature levels (3) Fully-connected Fusion: the complementary path is augmented to enrich feature for each proposal. 3> Performance: (1)3~4 point better thatn Mask-RCNN on COCO, Cityscapes. (2) 1st for instance segmentation and 2nd for detection on COCO. 4> disadvantages: (1) maybe slow. (2) computing expensive and memory consuming. 2. Pose2Seg:...

(1) put a figure and table in one row: https://tex.stackexchange.com/questions/6850/table-and-figure-side-by-side-with-independent-captions

1. Pascal VOC(eval_thre=0.5)