search

zhangliliang / RPN_BF

RPN+BF
Other
203 stars 75 forks source link

Question about machine #18

Closed SummerTrains closed 7 years ago

SummerTrains commented 8 years ago

First of all, thank you for your job! However, when I run script_rpn_pedestrian_VGG16_caltech.m , my computer jamed and can not do anything, staying in the stage "stage one RPN"! Then, I waited for about 1 hour, it's still jamed.
So I just want to ask the minimum requirement about machine. Here is my machine : One Titan x GPU; 32G memory. Is enough for this experiment? Thank you very much!

zhangliliang commented 8 years ago

Hi,

Actually the problem might lie on the CPU memory (not the GPU memory).

The RPN need to pre-compute and save all the anchors for all images, and store them in the memory during the training time.

The Caltech 10x dataset contains 4w+ images, and it might need a large memory.

Thus, it might help when reduce the training data, e.g. reduce to Caltech 3x or Caltech 5x.

SummerTrains commented 8 years ago

@zhangliliang Thank you! I'm reading your code. So which configuration can change Caltech 10x to Caltech 5x or 3x ? Can you tell me now ? Thank you!

zhangliliang commented 8 years ago

Hi,

Change the skip in the extract_img_anno.m.

Skip = 3 coresponds to caltech10x， skip = 6 corespond a to caltech5x.