jndeng / dacsdc-deepz Goto Github PK

I downloaded the samples1000.rar and deleted some images from that till I have 12 then it gave this error. Before that number of images was 23 and it worked. And I would like to find a way to print position of objects in each frame separately not taking whole detections into xml file. Is that possible?

Hi,

Should I do something extra for multiple object detection? Then, could you please guide me what it is ?

Thanks in advance,

Thanks for sharing the source code.

Where can I download the training data?

After gcc and nvcc some programs...it showed an error of "Multiple Definition" of linear_gradient..So strange...Do I have to do anything else before running make?

Is this a normal phenomenon? I mean, whether the reason it stops quickly is that I used the pretrained model you provided? Then how can I train the model from the very beginning?

In the contest main page, it is reported that your algorithm runs with 20 fps however I measured around 25 fps. Is that the reason because you included the timing for loading model weights or because I use TensorRT 4. If first statement is true as I see from your code includes that, it will converge into 25 fps given infinite length input. Could you please explain that ?