m-hamza-mughal / aerial-template-matching Goto Github PK

Hi, I want to know that did you train the network again using your own UAV datasets? and how do you label the train data? Or you test your data using the trained model in NCnet? Thanku

Can you publish the complete code in the paper "Assisting UAV Localization Via Deep ContextualImage Matching"? For example, the algorithm related to the experimental data in the article, and the code used to evaluate the results. I will use it for academic research. Thanks!

Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same or input should be a MKLDNN tensor and weight is a dense tenso

Hello,

Is there a trained model that you could provide (.pth.tar file)?

Also, would you be able to explain how the dataset is set up in further detail? It looks like the rows correspond to the images and the columns are sets of points, but when plotting the first row of points on Image1 and the orthomosaic, the sets of points don't correlate. I assumed the columns to be something like Image1_x1, Image1_y1, Orthomosaic_x1, Orthomosaic_y1, Image1_x2, Image1_y2, Orthomosaic_x2, Orthomosaic_y2, etc. I also tried variations on this, but haven't found anything that works yet.

Thanks in advance!

hi, I am wondering that why you choose to use fcn to regress the points in the last step