Ashish kumar

Co-Founder @ ScoreLabs AI
IIT Kanpur
Just Disruption, Nothing Else !!!

[Ashish kumar @ashishkr822]

@ccc_jdvvnl Please check your inbox

[Ashish kumar @ashishkr822]

@ccc_jdvvnl In pilibanga, ward no 6, there is no electricity for the past 11 hours in a particular lane. Have reached many local contact number of the electricity department however not even a single number os functional. It's sad to see such conditions in 2024. Please resolve.

[Ashish kumar @ashishkr822]

@CVPR
Want to speed up or downscale or dynamically prune ConvNets, Transformers without loosing accuracy?

Checkout our 🔥#CVPR2024 paper, Pick-or-Mix: Dynamic Channel Sampling for ConvNets

Drop by Poster: 103, June-19, 17:15 Arch-4AE
PDF and Code: https://t.co/9CtTvALiXH

[Ashish kumar @ashishkr822]

I am looking for Research position in autonomous vehicles/robotics/UAVs, neural network architecture design. I have publications in ICRA, IROS, RA-L, CVPR.

See: https://t.co/41yYtNlhJR

🔥I am in Seattle to present my paper in @CVPR 2024, Poster 103, 19 June, 17:15 Arch 4A-E

[Ashish kumar @ashishkr822]

@ieee_ras_icra Video: https://t.co/ASEQA3WuwC

[Ashish kumar @ashishkr822]

If you are at @ieee_ras_icra , tune in for our work "High-Speed Stereo Visual SLAM for Low-Powered Devices.
#ICRA2024

Session: Aerial Systems: Perception and Autonomy I

Paper: https://t.co/WQzd7jUjrK
Code: https://t.co/pbchwuic6o

[Ashish kumar @ashishkr822]

We need to be serious for the rising cancer and kidney patients in India. It's a massive challenge for our country's growth and has devastating aspects. E.g. youngsters loaded with responsibilities of their loved ones when they can do a substantial work for country's development.

[Ashish kumar @ashishkr822]

@chriswolfvision @iclr_conf @CVPR @NeurIPSConf @icmlconf ICLR establishing new bars of high standard !!!

[Ashish kumar @ashishkr822]

@chriswolfvision @CSProfKGD what a coincidence !!...so true😅😂😂

[Ashish kumar @ashishkr822]

@YiMaTweets It is the tag line of top research venues (i won't name) on their application portal that "applicant must have published in those so-called top conferences". Without that, I don't think one'll get a reply. This discourages me to apply until I publish in those. Sad it is.

[Ashish kumar @ashishkr822]

@Ken_Goldberg @michellehampson @IEEESpectrum @AUTOLab_Cal @IntuitiveSurg @UCBerkeley impressive work professor !!!... As per the video, a laser based 3D sensing has been used. But during surgery, the instrument is inside the body, in that robot will have to rely only on the video feed. So i think, there is a lot more room of work on this problem. What do you say?

[Ashish kumar @ashishkr822]

@EngNadeau @Ken_Goldberg Yes you are right..!!!.. But I have encountered many of such..!!!...And i feel wish i could convince them in person!!..

[Ashish kumar @ashishkr822]

@EngNadeau @Ken_Goldberg but then there are reviewers who say "hey your method fails on this. How can it be state-of-the-art". So It is really important that reviewers also change their perception about "improvements, failures and novelty".

[Ashish kumar @ashishkr822]

impressive results!!!

[Xiaowei Zhou @XiaoweiZhou5]

No CAD models, no instance or category-specific training, all you need is a strong generic feature matcher like SuperGlue or LoFTR. Check out our CVPR 2022 paper on object pose estimation without CAD models!

[Ashish kumar @ashishkr822]

@YiMaTweets I also faced same issue in #CVPR2022

[Ashish kumar @ashishkr822]

@CSProfKGD Dear Prof. Why it is difficult to find reviewers ? Is it happening this time only or is it as always ?

[Ashish kumar @ashishkr822]

@jbhuang0604 @CSProfKGD I love the titles.. omg...!!

[Ashish kumar @ashishkr822]

@davsca1 Based on the quantitative metrics in the paper, the translation errors are really very low. It shows how much the sensor itself can affect an algorithm.

[Ashish kumar @ashishkr822]

@antoniloq @davsca1 I was planning to generate commands in unsupervised manner while taking the obstacle information explicitly. But both the functions provides these information independently not simultaneously.