Hello, we are the master students in National Yang Ming Chiao Tung University.
We have a few questions after reading the Deep Object Pose Estimation for Semantic Robotic Grasping of Household Objects. In the paper, the figure 4 showed that the speed would increase while the stages increasing. The question we want to ask is that, what is “speed” refer to. Besides, why the speed would increase during stages 1 to 6, while the model structure and parameters were still the same?
The following picture is from the paper.
Best regards,
Shao-Ning Lu, Sheng-Ye Wen, Yuan-Chieh Tsai.
Thanks for reading the paper ([1809.10790] Deep Object Pose Estimation for Semantic Robotic Grasping of Household Objects) - and also thanks for taking the initiative to post a question here on our forum.
As you can expect our researchers are pretty busy - but Stan is a great guy, so I have send him an email asking for a comment.
Wishing you the best for your studies!
Each stage contains several layers. So the number of stages is related to the depth of the network. “Speed” is really “computing time”. Therefore, the computing time increases as you go down the table, because the network gets bigger.
