How to use image-wise (IW) and object-wise (OW)Cornell data splits mentioned in the paper?Did you use data augment by default when you trained Cornell?Thank you.

dougsm / ggcnn

Generative Grasping CNN from "Closing the Loop for Robotic Grasping: A Real-time, Generative Grasp Synthesis Approach" (RSS 2018)

BSD 3-Clause "New" or "Revised" License

484 stars 138 forks source link

Closed thk8181 closed 2 years ago

dougsm commented 2 years ago

Hello, The code here only supports image-wise split, which is what's quoted in the paper. The data is augmented by default, with random crop and rotate (https://github.com/dougsm/ggcnn/blob/master/train_ggcnn.py#L202)

thk8181 commented 2 years ago

Thank you for reply.