Making Your Python Code Run Faster.
Brandon Rohrer. https://sigmoid.social/@brohrer@recsys.social .

pdf: https://tyr.fyi/6
code: https://github.com/brohrer/how-to-train-your-robot/tree/main/chapter_6
web: https://tyr.fyi

#python #optimization #AmyPostspapers

Resilient and Distributed Multi-Robot Visual SLAM: Datasets, Experiments, and Lessons Learned
Yulun Tian, Yun Chang, Long Quang, Arthur Schang, Carlos Nieto-Granda, Jonathan P. How, Luca Carlone

abs: https://arxiv.org/abs/2304.04362
code: https://github.com/MIT-SPARK/Kimera-Multi
data: https://github.com/MIT-SPARK/Kimera-Multi-Data

#Robotics #SLAM #arXiv #AmyPostsPapers #hasCode

Remember SAM from < week ago? Here's the 1st arXiv reference I have seen to it, a tech-report style paper.

Can SAM Segment Anything? When SAM Meets Camouflaged Object Detection
Lv Tang, Haoke Xiao, Bo Li

abs: http://arxiv.org/abs/2304.04709
results stub: https://github.com/luckybird1994/SAMCOD
#arXiv #ComputerVision #Segmentation #AmyPostsPapers

From Zero to Hero: Convincing with Extremely Complicated Math
Authors: Maximilian Weiherer, Bernhard Egger

The Association for Computational Heresy conf. 23.

abs: https://arxiv.org/abs/2304.00399
code: https://github.com/mweiherer/zero2hero

This is a joke paper, but it also says something about the research communities involved, yes?

#arXiv #AmyPostsPapers

Segment Anything
Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollár, Ross Girshick

abs: https://buff.ly/3zDpNp6
web, demo, dataset: https://buff.ly/3nI929j
code on GH: https://buff.ly/3zAPG8P

The demo is very cool! Video alt: Interactive prompting for segmentation, photo of a snail on top of frog, on top of turtle.
#arXiv #ComputerVision #Segmentation #AmyPostsPapers

Photometric LiDAR and RGB-D Bundle Adjustment
Luca Di Giammarino, Emanuele Giacomini, Leonardo Brizi, Omar Salem, Giorgio Grisetti

tl;dr: After initial SLAM estimate, refinement w/ this Photometric BA method.

LiDAR and depth data doesn't have color, define photometric to mean comparing depth on image1 & transformation(image2).

Very good Related Work section.

#Robotics #SLAM #arXiv #AmyPostsPapers
abs: http://arxiv.org/abs/2303.16878
GH (stub for now): https://github.com/digiamm/ba_md_slam

Persistent Nature: A Generative Model of Unbounded 3D Worlds
Lucy Chai, Richard Tucker, Zhengqi Li, Phillip Isola, Noah Snavely

Input -- single-view landscape photos, output -- consistent world model for simulating flights through the scene.

#arXiv #AmyPostsPapers #GenerativeWorlds
abs: https://arxiv.org/abs/2303.13515
web: https://chail.github.io/persistent-nature/
video alt: flyover of a region with low mountains and streams.

@oliversampson This is a super interesting field! I was reminded of a paper from my arXiv list that deals with some of these topics --

Marginalia and machine learning: Handwritten text
recognition for Marginalia Collections
Adam Axelsson, Liang Cheng, Jonas Frankem ̈olle and Ekta Vats

#AmyPostsPapers
abs: https://arxiv.org/abs/2303.05929

Next-Best-View Selection for Robot Eye-in-Hand Calibration
Jun Yang, Jason Rebello, Steven L. Waslander

tl;dr: in calibration, how do we know we have a good dataset? Uses a next-best view policy to reduce the # of poses & errors in robot-camera context.

(btw, I have been asked about doing something like this in every talk I give on robot-camera calibration. Yeah! someone has already done it.)

abs: http://arxiv.org/abs/2303.06766
#arXiv #AmyPostsPapers #RobotCameraCalibration

Deep-Learning-based Counting Methods, Datasets, and Applications in Agriculture -- A Review
Guy Farjon, Liu Huijun, Yael Edan

Counting is imporant to agri. for crop load prediction, spray, estimating labor needs.

abs: https://arxiv.org/abs/2303.02632
#arXiv #Agriculture #review #AmyPostsPapers

Visual Place Recognition: A Tutorial
Stefan Schubert, Peer Neubert, Sourav Garg, Michael Milford, Tobias Fischer

[15 pages] presents VPR for newcomers & deals with evaluation. Maybe the code would serve as a good baseline method?

abs:https://arxiv.org/abs/2303.03281
code: https://github.com/stschubert/VPR_Tutorial
#VPR #ComputerVision #arXiv #AmyPostsPapers

Multi-View Reconstruction using Signed Ray Distance Functions (SRDF)
Pierre Zins, Yuanlu Xu, Edmond Boyer, Stefanie Wuhrer, Tony Tung

tl;dr: Signed Ray Distance Function (for spatial consistency) and a photo-consistency function are maximized over a volume. Lots of comparisons to learned and classical methods. Combines implicity parameterization with a focus on recovering geometry.

abs: https://arxiv.org/abs/2209.00082
code: upon accept.
#arXiv #computerVision #multiViewReconstruction #AmyPostsPapers

General, Single-shot, Target-less, and Automatic LiDAR-Camera Extrinsic Calibration Toolbox
Kenji Koide, Shuji Oishi, Masashi Yokozuka, Atsuhiko Banno

ICRA2023

tl;dr: uses SuperGlue or a GUi to create correspondences needed for LiDAR<->Camera calibration.

#arXiv #AmyPostsPapers #robotics #calibration

abs:
https://buff.ly/3lAwgx2
prelim code, in the devel branch:
https://buff.ly/3XtuHOL

Ten Lessons We Have Learned in the New "Sparseland": A Short Handbook for Sparse Neural Network Researchers
Shiwei Liu, Zhangyang Wang

What a beautifully-written, kind, and academically-funny abstract!

favorite part: "At the very least ... if you are writing/planning to write a
paper or rebuttal in the field of SNNs, we hope some of our answers could help
you!

abs: https://arxiv.org/abs/2302.02596
pdf: https://arxiv.org/pdf/2302.02596.pdf

#arXiv #SNNs #AmyPostsPapers

Comparison of modern open-source visual SLAM approaches
Sharafutdinov et al.

tl;dr: title, comparative analyis of algos and datasets, Docker images!

abs:
https://arxiv.org/abs/2108.01654
GH w/ Docker info:
https://github.com/KopanevPavel/SLAM-Dockers
#SLAM #ComputerVision #Robotics #arXiv #AmyPostsPapers

Few-Shot Learning Enables Population-Scale Analysis of Leaf Traits in Populus trichocarpa
Lagergren et al.

tl;dr: leaf phenotyping methods (using few-shot learning) applied to Black Cottonwood tree -- traits are leaf shape and veins. GWAS applied to the traits. Beautiful figures!

arXiv: https://arxiv.org/abs/2301.10351
GH: https://github.com/jlager/few-shot-leaf-segmentation
data: https://doi.ccs.ornl.gov/ui/doi/415
#ComputerVision #arXiv #PlantPhenotyping #AmyPostsPapers

The 8-Point Algorithm as an Inductive Bias for Relative Pose Prediction by ViTs

Chris Rockwell, Justin Johnson, David F. Fouhey

accepted 2022.

Baseline for learned relative pose estimation, using a modified ViT.

#arXiv #AmyPostsPapers #ComputerVision

abs: https://arxiv.org/abs/2208.08988
GH: https://github.com/crockwell/rel_pose
web: https://crockwell.github.io/rel_pose/

@ducha_aiki Open access conference paper version (I don't have access to IEEE xPlore)

#AmyPostsPapers

https://openaccess.thecvf.com/content/ICCV2021/html/Lindenberger_Pixel-Perfect_Structure-From-Motion_With_Featuremetric_Refinement_ICCV_2021_paper.html

@eric_brachmann @ducha_aiki @mihaidusmanu

It is happening! Research on the M-site. I now need to read a bit to understand conversations.

Paper ref. for LaMAR
LaMAR: Benchmarking Localization and Mapping for Augmented Reality

Paul-Edouard Sarlin, Mihai Dusmanu, Johannes L. Schönberger, Pablo Speciale, Lukas Gruber, Viktor Larsson, Ondrej Miksik, Marc Pollefeys

ECCV 2022

abs: https://arxiv.org/abs/2210.10770
website: https://lamar.ethz.ch/

#AmyPostsPapers #AR #Benchmarks

maplab 2.0 -- A Modular and Multi-Modal Mapping Framework

Andrei Cramariuc, Lukas Bernreiter, Florian Tschopp, Marius Fehr, Victor Reijgwart, Juan Nieto, Roland Siegwart, Cesar Cadena

Open-source project that incorporates disparate data -- camera, GPS, IMU, LIDAR, multi-robot. ❤️ icons in Table II (3rd image).

#AmyPostsPapers #arXiv #ComputerVision #Robotics #SLAM
abs: https://arxiv.org/abs/2212.00654
ieee: https://ieeexplore.ieee.org/document/9976204
GH: https://github.com/ethz-asl/maplab