steven konev

Machine Learning Engineer

[steven konev @KonevSteven]

StandardE2E 0.0.7 adds KIT MRT's KITScenes Multimodal & LongTail datasets plus a universal visualizer: any processed dataset to video, cameras + BEV, no per-dataset code.
https://t.co/KX0M3iqFBO

[steven konev @KonevSteven]

StandardE2E 0.0.6 adds three datasets: @torc_robotics's TruckDrive, @tudelft's View-of-Delft, and @motionaldrive's nuScenes — cameras, LiDAR, 3D boxes & HD maps, all through the same processing and unified PyTorch DataLoader.
https://t.co/KX0M3iqFBO

[steven konev @KonevSteven]

StandardE2E now supports @comma_ai 's comma2k19 dataset from 0.0.5 version: ~2,000 one-minute highway segments — front camera + ego trajectories — now load through the same processing and unified PyTorch DataLoader.
https://t.co/1RCdvkk37j

[steven konev @KonevSteven]

Sharing what I've been working on: StandardE2E 🚗A unified API + PyTorch DataLoader for end-to-end AV datasets:
@Waymo (E2E + Perception), Argoverse 2, NAVSIM &
@wayve_ai Scenes101 and more WIP, all in one format. Less data plumbing, more model building!
https://t.co/KX0M3irdrm

[steven konev @KonevSteven]

In https://t.co/pO80v2psio, we build a citizen science platform and autonomous boats that collect rich data about urban water quality. But collecting data is only the first step.
With @googlegemma 4, we transform it into compact, informative, and interactive actionable reports.

[steven konev @KonevSteven]

As part of the https://t.co/5oh46s3yV4 project, we developed a small, affordable autonomous boat that, despite its size, can travel for miles and continuously collect water quality data. For more details, see the write-up: https://t.co/OIX4dv38UV

[steven konev @KonevSteven]

Just shipped something tiny I love💧
https://t.co/5oh46s3yV4 — spot something off in an Amsterdam canal? Send a photo to a Telegram bot. #AI reads the photo, classifies the issue, pins it on a live public map. No app, no signup
#Amsterdam #grachten #citizenscience #water #ecology

[steven konev @KonevSteven]

Excited to share that I won a silver medal 🥈 at Kaggle Stanford Ribonanza RNA Folding competition 🧬(<-it is not RNA I know:) ! Thanks to all the organizers and participants for this enriching experience and the great opportunity!https://t.co/PE2ULziwzS

[steven konev @KonevSteven]

Lately I became a prize winner of AISG-SLA Visual Localisation Challenge organized by @AISingapore with solutions presented at @IJCAIconf 23!
Announcement with brief solutions desc https://t.co/TtM6ArMNdH

[steven konev @KonevSteven]

21 [MotionCNN] https://t.co/lrD2CZez8N
22 [MPA++] https://t.co/VPwsR89Tf2

[steven konev @KonevSteven]

This year @Waymo hosts Motion Prediction Challenge for the 3rd time. If you are a fan of autonomous driving you defenitely should participate! For your smooth start I share my prize winning solutions from previous years
⭐️ 21 https://t.co/BTksn55fDL
⭐️ 22 https://t.co/YQyixUcjys

[arXiv Daily @Arxiv_Daily]

MPA: MultiPath++ Based Architecture for Motion Prediction
https://t.co/Wyqb1P8SkS
by @Konevsteven
#ComputerVision #PatternRecognition

[Artsiom Sanakoyeu @artsiom_s]

We've finally uploaded our report with winning solution at CVPR21 Autonomous Driving Workshop to arxiv!

[steven konev @KonevSteven]

The code of the solution https://t.co/VPwsR89Tf2

[steven konev @KonevSteven]

I am proud to announce that this year I achieved the 3rd result on @Waymo Motion Prediction Challenge leaderboard (as of May, 26) and Honorable Mention.
The report is available on @CVPR Workshop on Autonomous Driving https://t.co/xEoVhXnfaE
The code is available on GitHub thread

[steven konev @KonevSteven]

@BorisJohnson This war is NOT in our name. Thank you for understanding this, sir!

[Jia-Bin Huang @jbhuang0604]

Somehow my timeline is filled with papers where the models were trained with 64 A100 GPUs for weeks.

Grad students with two 2080Ti:

[Artsiom Sanakoyeu @artsiom_s]

I'm happy to announce that our team (me, @KonevSteven, K. Brodt) was awarded 3rd place within the Waymo Motion Prediction Challenge 🥳

Task: predict trajectories of the agents for 8 seconds into the future.

📜Technical report https://t.co/LRxrgCibta
We also released our code ↓

[Artsiom Sanakoyeu @artsiom_s]

Waymo motion prediction challenge 2021: 3rd place solution.

⚒️Code https://t.co/dgrO61pq0v