search

SeokjuLee / VPGNet

VPGNet: Vanishing Point Guided Network for Lane and Road Marking Detection and Recognition (ICCV 2017)
MIT License
487 stars 166 forks source link

How to evaluate the results #12

Open daixiaogang opened 6 years ago

daixiaogang commented 6 years ago

I am now doing the evaluation of the results ,in your paper you said: we compute the minimum distance from the center of each cell to the sampled lane points for every cell. If the minimum distance is within the boundary R, we mark these sampled points as true positive and the corresponding grid cell as detected. how about the R and the class FP?

SeokjuLee commented 6 years ago

@daixiaogang We set 'R=20' to equal to the average value of approximately half of the lane thickness (paper section 5.4).

daixiaogang commented 6 years ago

@SeokjuLee ,Thanks for your reply . And I wonder I to calculate the center of the grid,as it is important to the FP? In my code, I use the (xmin+4,ymin+4) as the center and use the euclidean metric to calculate the distance between predict and GT,am I right?

SeokjuLee commented 6 years ago

@daixiaogang What is the FP? With the grid size of 8, the center position you defined is right.

SkyeFromKindergarten commented 6 years ago

FP means False Positive,commonly called a "false alarm". It is more like that there is no such an object and somehow it has been detected as an object. For example, there's no car in the road and the output tells you that there's a car.

daixiaogang commented 6 years ago

@SeokjuLee ，can you release your evaluate code on caltech lanes detection?

ryan1051 commented 5 years ago

@daixiaogang Hi, I trained a model. Now I want to test a image with the model, can you show me how to evaluate it?