Top Tweets for #hendrikspapers
Very cool work at #ECCV2020 by Zhengqi Li et al about "Crowdsampling the Plenoptic Function" which allows reconstruction of multi-plane images+latent code from crowd-source images for rerendering under varying illumination. 1/3
#lightfield #hendrikspapers @Jimantha
Who has not used, tweaked (and then rewritten from scratch) some adaptive binarization method! Very nice too see classic approaches being extended and generalized! Also a very nice and concise video. #ECCV2020 #hendrikspapers
I got really into a bunch of 1980s-era papers about histogram thresholding, and wrote a weird paper. I sent it to ECCV assuming the reviewer response would be "why are you writing a direct response to two 40 year old papers" but they loved it, so hey. https://t.co/BgZ9dETc8p
Yep, two very good reads, one old classic (plenoptic function) and one possibly becoming a modern one (NeRF).
#hendrikspapers
The classic “The Plenoptic Function and the Elements of Early Vision” by
Adelson and Bergen from 1991 is fun reading for those studying view synthesis with neural radiance fields. #computervision #graphics #ECCV2020
https://t.co/MjP0nsaWJj
NeRF paper:
https://t.co/6UtVfqYejZ
For all those fighting to get their batch size up to one. This (getting rid of normalization layers) could have some huge practical advantages!
#hendrikspapers
How to train very deep ConvNets without residual blocks? Our ICML paper on Deep Isometric Learning successfully trains 100-layer ConvNets without any shortcut connections nor normalization layers (BN/GN) on ImageNet.
Paper: https://t.co/mxMJO1Wub9
Code: https://t.co/8Omd8SYYS9
#CVPR2020 pages are still online. Like Michal Irani talking at the Workshop for Computational Cameras and Displays about "Deep Internal Learning", without external (to the test image) training data, as well as the use of deep linear networks (link see comment!)
#hendrikspapers
Very interesting panel discussion at the Workshop on Autonomous Driving panel discussion at #CVPR2020!
https://t.co/IU9vBp1w0k
Many thanks to all the speakers. Brief summary in the replies, hope to see the full thing on youtube!
#hendrikspapers #CVPR #cvpr20 #AutonomousVehicles
Very refreshing talk at the #CVPR2020 Workshop on Autonomous Driving by Andreas Wendel from @KodiakRobotics. Talking about their autonomous truck operations with lots of hands-on examples:
https://t.co/hrhpaZIbPJ
#hendrikspapers #cvpr20 #CVPR
Very nice talk by Deva Ramanan about vision for autonomous driving. Interesting thoughts about accuracy vs latency in dynamic scenarios and more generic "natural" perception (things vs classes).
https://t.co/e2b9CCcJxn
#hendrikspapers #autodriving #AISafety @roboVisionCMU
Cool work using continuous representations. Logical next steps: Blockwise operation, overlapping blockwise operation, wavelets ...
Link to project page: https://t.co/5HXfcJdWqH
#hendrikspapers
A non-linearity that works much better than ReLUs. The work described in this video might also be relevant to understanding grid cells.
https://t.co/YlV9j79fgi
A reminder that overfitting can happen in many ways: "Don’t Judge an Object by Its Context: Learning to Overcome Contextual Bias" by Krishna Kumar Singh et al explicitly handles contextual bias during training, greatly improving generalization!
#hendrikspapers #CVPR2020 @deeptigp
Very nice approach by @zamir_ar et al, exploiting "Robust Learning Through Cross-Task Consistency" improving various predictions in an automatic and principled fashion. With live demo - showing my desk in preparation for #CVPR2020 "night" sessions (I'm UTC+2).
#hendrikspapers
And another approach to semantic mask refinement: @kirillov_a_n et al show "PointRend: Image Segmentation As Rendering". This work implements pointwise refinement. Improves SOTA , plug & play on top of any FCN! Awesome!
#hendrikspapers #CVPR2020 #CVPR @inkynumbers
Scalespace is up and alive. Nice high resolution semantic masks: Ho Kei Cheng et al show "CascadePSP: Toward Class-Agnostic and Very High-Resolution Segmentation via Global and Local Refinement", using patch-wise refinement to calculate ultra crisp masks. Neat!
#hendrikspapers
Cool new stuff: Dahun Kim et al. introduce "Video Panoptic Segmentation" a straightforward extension demanding instances to stay consistent over time! Includes dataset extensions, metrics, and solutions, also push previous SOTA! Very Neat!
#hendrikspapers #CVPR2020 #CVPR
Very nice work by Ivan Anokhin et al. showing "High-Resolution Daytime Translation Without Domain Labels", demonstrating high quality daytime transfers using GANs and ... training. Seriously, nice results, and shows why GAN design can be ... fiddly!
#hendrikspapers #CVPR2020
I need to catch up at flow SOTA. @simon_niklaus and Feng Liu show "Softmax Splatting for Video Frame Interpolation" which demonstrates super smooth and stable flow interpolation. Awesome results!
#hendrikspapers #CVPR2020 #CVPR
Who doesn't love face relighting: Nestmeyer et al. show in "Learning Physics-Guided Face Relighting Under Directional Light" how to incorporate directional lighting for more realistic relighting. Yay for better modeling!
#hendrikspapers #CVPR2020 @jflalondeqc @iainmatthews @ASM_L
Very cool approach by @xiaohangzhan et al. performing "Self-Supervised Scene De-Occlusion". Principled formulation and neat solution on how to create de-occlusion training data from instance segmentation labels.
#hendrikspapers #CVPR2020 #CVPR
Intersting variant of superresolution: Yuval Bahat and Tomer Michaeli implement "Explorable Super Resolution", where the user can locally influence the superresolution results while keeping the reconstruction consistent! Very nice!
#hendrikspapers #CVPR2020 #CVPR
Magical results by Xiaoming Li et al. using "Enhanced Blind Face Restoration With Multi-Exemplar Images and Adaptive Spatial Feature Fusion". Surprisingly stable even in video mode! Nice work!
#hendrikspapers #CVPR2020 #CVPR
Last Seen Hashtags on Sotwe
Th3CATS
Seen from United States
latino
Seen from United States
下种
Seen from United States
BitfinexLATAMTour
Seen from United States
gebzetraveſti
Seen from Turkey
uglybastard
Seen from Denmark
CAPCAconference2021
Seen from France
Religious
Seen from United States
minichat() +filter:native_video
Seen from Turkey
MaritalBreakthrough
Seen from Brazil
Trends for you
Most Popular Users
Elon Musk 
@elonmusk
241.7M followers
Barack Obama
@barackobama
118.9M followers
Cristiano Ronaldo
@cristiano
114.6M followers
Donald J. Trump
@realdonaldtrump
111.9M followers
Narendra Modi
@narendramodi
107.2M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.4M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90.1M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.5M followers
Kim Kardashian
@kimkardashian
70.9M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.5M followers
Bill Gates
@billgates
65.2M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.3M followers
CNN
@cnn
61.8M followers
X
@x
60.7M followers