Mohit Shridhar

@jcjohnson can open-world object detection be done in a similar way? Also, how do you use the boxes and features from `extractFeatures` in `self.nets.language_model:forward()` (the function expects image vectors and...

Unfortunately no. At least not in the near future. Sorry.

@yang9004 If I remember correctly, each room is represented with a square grid of 1m x 1m units with the origin somewhere in the center of the room. `-5|-6` correspond...

@DeeDive This was a weird choice I made during early experimentation. I thought asking the agent to generalize to all unseen colors would be too hard. So I mixed in...

Yeah the shared colors give an unfair advantage. This is only issue with colors though; shapes and Google Scanned Objects have no overlap. I think you can modify the color...

@emigmo, you can't. The code only works for batch-size=1. You would need to refactor the whole repo to support this.

Hi @siddk, There are a number of non-transporter baselines in the original [TransporterNets repo](https://github.com/google-research/ravens/tree/master/ravens/agents). But you have to reimplement them in PyTorch. The ConvMLP baseline by Zeng et al. might...

Closing due to inactivity. For future reference, see thread above.

Closing due to inactivity. For future reference, see thread above.

Thanks @alopezrivera! [❤️](https://emojipedia.org/red-heart/) Will look into this soon.