Shu Kong

Assistant Professor at UMacau; research affiliate at OIST; previously at TAMU, CMU, UCI;
a researcher in CV/ML & Interdisciplinary Research

[Shu Kong @AimeryKong]

@wjscheirer https://t.co/dfvKrRC1ji

[Shu Kong @AimeryKong]

@CVPR @wjscheirer @saakur3 @NeeharPeri https://t.co/dfvKrRC1ji

[Shu Kong @AimeryKong]

starting our CVPR workshop of Open World Vision 🍻 Time: 08:30-12:15 MDT (Denver Time), Room 712! @CVPR @wjscheirer @saakur3 @NeeharPeri

[Shu Kong @AimeryKong]

starting our CVPR workshop of AutoExpert, developing methods to automatically annotate domain-data with domain-expert crafted guidelines! 🍻 Time: 13:10-16:30 MDT (Denver Time), Room 711! @sarameghanbeery @CVPR
https://t.co/LKmupBib6N

[Shu Kong @AimeryKong]

will be attending the 43rd Mid-Continent Paleobotanical Colloquium in Panama and teaching a short course; excited to meet folks @smithsonian
https://t.co/I4s3KpNDpN

[Shu Kong @AimeryKong]

@CVPR There are two challenges. Please consider to play around them!

1. Auto-Annotation of Animal Species for Ecological/Biological Research (https://t.co/stSt99O4S6)

2. Auto-Annotation of 3D Cuboid for Autonomous Driving (https://t.co/jPPR35CC9D)

[Shu Kong @AimeryKong]

excited to meet folks @CVPR workshop of AutoExpert - Auto-Annotation from Expert-Crafted Annotation Guidelines https://t.co/LKmupBib6N

[Shu Kong @AimeryKong]

excited to meet folks @CVPR workshop of Open-World Vision https://t.co/dfvKrRC1ji

[Shu Kong @AimeryKong]

look forward to exciting demos @CVPR
https://t.co/1qlGSb0ASA

[Shu Kong @AimeryKong]

@ICCVConference excited to share "a little tweak" of prompt tuning - handling the "burstiness" and non-Gaussian tokens via whitening significantly accelerates learning and boosts performance. The final solution is a simple low-rank bilinear prompt tuning
https://t.co/YWTcKx9PNW

[Shu Kong @AimeryKong]

@ICCVConference excited to share a "little tweak" in prompt tuning - handling the "burstiness" and non-Gaussian of tokens through simple whitening significantly accelerate learning and boosts performance on downstream tasks
https://t.co/YWTcKx9PNW

[Shu Kong @AimeryKong]

The 5th Workshop @CVPR of Visual Perception and Learning in the Open World will start at 9am CDT, June 11! Look forward to exciting talks and challenge competitions!
https://t.co/3YfqhllySN

[Shu Kong @AimeryKong]

comparing DL-quantized pollen traits & environmental factors reveals abundant results: temp & solar radiation influence pollen morphology; thermal stress drives an increase in pollen size; higher ultraviolet B radiation selects for thicker corpus walls... https://t.co/WjpLSf4QEP

[Shu Kong @AimeryKong]

check out @CVPR paper on realistic few-shot recognition motivated by a desire of replacing human annotators with foundation models in a specific task which provides annotation guidelines that contain few-shot visuals and test descriptions
https://t.co/FZObdhrSW8

[Shu Kong @AimeryKong]

check out our @CVPR paper that solves Instance Detection in an open world perspective -- embracing the open world, its foundation models and open data to boost performance!

https://t.co/HOSjOnwIr1

[Anish Madan @anishmadan23]

Visit our #NeurIPS2024 poster today to learn more about leveraging Foundational VLMs for generating high-quality annotations!

Location: East Exhibit Hall A-C, #3407
Time: 4:30–7:30 PM (PST)

[Shu Kong @AimeryKong]

Congrats to Dr. @MilefNicholas, my first PhD student, on the successful PhD defense and the new role as a Research Scientist at @Meta Can't believe that we haven't met in person at all after years of remote collaboration:-)

[Shu Kong @AimeryKong]

Congrats to Dr. @Marc_Adaime on the successful PhD defense and the new role as Research Fellow @smithsonian
I really enjoy working together in the past years with @cyberpollenlab Can't believe that we haven't met in person over >10 years of collaboration:-)

[Shu Kong @AimeryKong]

#ECCV2024 come to our booth 222 @eccvconf today (Oct 4, Friday morning) to talk about how shoeprint matching is used in the super sensitive areas - criminal investigations, forensic examination, criminal justice, etc. Paper below:
https://t.co/6orX7RP14D

[Shu Kong @AimeryKong]

For safe driving with LiDAR perception, let's consider "Lidar Panoptic Segmentation in an Open World" to attention to both unknown and known things and stuff! See our IJCV work by @anirudhchak Meghana, @peiyunhu @lealtaixe @RamananDeva @AljosaOsep

https://t.co/DBGQWNT6IT