How is 6.7 million dataset generated from 1500 objects - the provided database generate around 1 million only

BerkeleyAutomation / dex-net

Repository for reading the Dex-Net 2.0 HDF5 database of 3D objects, parallel-jaw grasps, and robust grasp metrics

https://berkeleyautomation.github.io/dex-net/code.html

Other

296 stars 97 forks source link

How is 6.7 million dataset generated from 1500 objects - the provided database generate around 1 million only #74

Closed aprath1 closed 2 years ago

aprath1 commented 2 years ago

Does any one know how 6.7 million dataset is generated using dexnet 2,0 database ? dexnet 2.0 database - https://berkeley.app.box.com/s/eaq37px77jxktr8ggti016pr3gpudp9l using the generate database code : https://github.com/BerkeleyAutomation/dex-net/blob/master/tools/generate_gqcnn_dataset.py The resulting dataset is around 1 million only, so how is 6.7 million dataset generated? Does it mean the same 'generate_gqcnn_dataset.py' script is run on the same dexnet 2.0 database multiple times?

kondrauer commented 2 years ago

You can alter images_per_stable_pose in cfg/tools/generate_gqcnn_dataset.yaml to get more datapoints. Its default is 5 if you choose a higher number you will get more.