somuSan
@somuSan_
Followers
865
Following
15K
Statuses
2K
Sr. Computer Vision Engineer @OpenCVUniverse | xAutonomous Vehicle Research Assistant @IIITDelhi, 4x @kaggle expert, Former ML intern @IITKgp, @UNSW
BLR
Joined February 2020
Our team got a bronze🥉on the recent HuBMAP #kaggle competition We pulled a lot of hard work but were still stuck at 0.79 and missed the silver And this🥉made me a comp expert/4x expert🍻 We wanted to scale up at the end & @jarvislabsai,@vishnuvig was really helpful,tysm🙇‍♂️ 1/n
5
2
48
RT @swayaminsync: 🚀 Big milestone alert! "My book Concurrent C++: A Practical Guide" just hit 10 completed chapters, covering the fundamen…
0
1
0
@bipashasen31 @tangiblerobots this is crazy! i have been following your journey, and what you have done is so inspiring and going to motivate other ppl. LFG!
0
1
2
@jino_rohit im not using but want to using. but codeforces is the best. generally ppl go to A2OJ ladder and solves problem based on the rating of qsn they wanna practice (starts with <1300). And at the same time participate on div2/div3 contests. if you are also doing it, let me know.
1
0
1
Best video I saw today. How randomization can help make computations faster.
0
1
6
RT @jonstephens85: Do you want 3DGS and rendered videos FAST!? Than this is a tutorial for you! I reimplemented NVIDIA's InstantSplat to w…
0
12
0
Will be live in an hour, going through how NeRF(Neural Radiance Fields) works, and and how to use it @opencvlive If u r in computer vision/graphics started to learn about NeRFs/3dgs u should definitely join it. Or if you missed it see the recording!
0
1
9
@antoinepirrone @Mankaran32 I knew something like this is going to come soon. Great work guys!
0
0
4
📢 Gaussian Splatting Blog Post Alert! In this article i deep dive in the paper "3D Gaussian Splatting as Markov Chain Monte Carlo". Use the 3DGS MCMC + Bilateral Grids for Floater-Free 3D model and mask guided bg subst 3dgs. Finally simple application insert objects in 3DGS👇
If you’ve been using 3D Gaussian Splatting (3DGS) to create 3D representations of an environment, you might be familiar with the challenge of "floaters," which can be very difficult to remove. In this article, we demonstrate how to train your 3DGS model using MCMC combined with a bilateral grid to achieve "floater-free" 3D models. We provide a detailed explanation of the paper "3D Gaussian Splatting as Markov Chain Monte Carlo," covering all the equations and derivations step by step. Additionally, we show how to seamlessly insert small objects into a large 3D Gaussian environment using techniques like background segmentation and Gaussian manipulation.
0
1
19
RT @LearnOpenCV: If you’ve been using 3D Gaussian Splatting (3DGS) to create 3D representations of an environment, you might be familiar wi…
0
33
0
RT @janusch_patas: I've started experimenting with volumetric videos. Although currently unoptimized, I'm able to train (hopefully) arbitra…
0
7
0
RT @shishirny: This has been more than two years in the making. With the help of @artparkindia and generous funding from the government, we…
0
241
0
If you ever wondered why Gaussian Splatting has so many spherical harmonics components, here is your answer.
0
0
8
Tried running 3d convex splatting on a 16 GB vRAM system, and within 800 iteration it threw OOM. Ppl seem to be impressed by how good it is, but it comes with a significant cost.
0
0
3