Excuse me, during my training of your D2Det instance model with iSAID dataset, cuda out of memory at 800 iterations. I cropped the original images into 800×800 pixels with stride 200 and used 1 Titan GPU with a batchsize of 2. Could you please give me some advice ?
Excuse me, during my training of your D2Det instance model with iSAID dataset, cuda out of memory at 800 iterations. I cropped the original images into 800×800 pixels with stride 200 and used 1 Titan GPU with a batchsize of 2. Could you please give me some advice ?