Ankur Handa comments

Results 22 comments of


                                            Ankur Handa

Adding a new feature for reward sketching

@makseq I'd like to draw a continuous function with X-axis being the frame number and Y-axis being the reward. To clarify what that means check the reward sketching section in...

Adding a new feature for reward sketching

Do you think it is worth adding that in label-studio? I implemented it in python already (using matplotlib) but it's fairly simple and often requires more clicking.

Adding a new feature for reward sketching

The code looks like this in case you're interested. ` import math import numpy as np import matplotlib.pyplot as plt from matplotlib.backend_bases import MouseEvent class DraggablePlotExample(object): u""" An example of...

Adding a new feature for reward sketching

some of this is inspired from https://github.com/yuma-m/matplotlib-draggable-plot/blob/master/draggable_plot.py

Adding a new feature for reward sketching

Yes, sure. Let me know if you ever get to it. I'm happy to do testing for you. Thanks a lot for listening.

Can I make a layer of which input is disparity & output is Rt?

you can take inspiration from this https://github.com/ankurhanda/gvnn/blob/master/imagewarpingSE3.lua

Where is the M-estimators?

could you use Huber here: https://github.com/viorik/extracunn/blob/master/Huber.cu I'm writing a tf versoin of this so will release soon. torch is too old.

Where is the M-estimators?

Sorry busy with lots of other stuff at the moment! Will let you know when I do.

Where is the M-estimators?

@vbalnt have you seen this https://stackoverflow.com/questions/39106732/how-do-i-combine-tf-absolute-and-tf-square-to-create-the-huber-loss-function-in

Demo or example for 3D data

Can you tell me exactly what you'd like to do with the voxel data? What sort of operation would you like to do on the voxel grid?