Keep Sharp

Domain adaptation with caffe features (Experiment)

Concepts

While most images for training are produced with high quality using digital cameras, maybe with label, the domain of interests in real world could differ in some combination of factors, including scene, intra-category variation, object location and pose..., maybe without label, domain adaption algorithms is trying to minimize this ...

Single View Metrology

This computer vision project is to construct a 3D model from a single image plus some prior knowledge about the objects' coordinates in the image. It's based on this paper. You can download the code from here. Here is a sample 3D model (right) built from a image (left ...

Dense Light Stereo

This computer vision project is to build 3D model of objects based on dense light images, code can be downloaded from here, for the description and relevant papers check README in the code, here only shows the results.

Smoother parameter \(\sigma\) can not be too large, either too small. The ...

Intelligent Scissor

This computer vision project is to realize the function of cutting out the objects you are interested inside a picture and then save it as another picture which has transparent background except the cut out objects. You can download the code from here. Here is an example of Lena picken ...