2023年8月のポストまとめ

📊 概要

20件のポスト（リプライ除く）を5カテゴリに分類しました。

カテゴリ	件数	割合
🏗️ 3D再構成・SLAM	13	██████████ 65%
🚗 自動運転	1	█ 5%
🤖 ロボティクス	2	██ 10%
📄 論文紹介	2	██ 10%
💬 その他	2	██ 10%

🏆 人気トップ3

🥇 1位

RT	36
Like	185
Views	10000

Fuzzy Metaballs+ use 3D Gaussians for differentiable rendering, integrating optical flow and producing watertight meshes. These techniques ensure speedy, reliable reconstructions on both GPU and CPU.

🔗 ポストを見る

---

🥈 2位

RT	23
Like	111
Views	6000

S-NeRF is a system that uses NeRFs to synthesize expansive and realistic street views for applications like autonomous driving simulation, robotics, and augmented reality. S-NeRF: Neural Radiance Fields for Street Views

🔗 ポストを見る

---

🥉 3位

RT	13
Like	76
Views	5700

Most Vision-and-Language Navigation methods use recurrent states, topological maps, or top-down semantic maps for memory. The new approach uses Grid Memory Map for spatial representation and emphasizes visual cues in each grid. GridMM

🔗 ポストを見る

---

📂 カテゴリ別ハイライト

🏗️ 3D再構成・SLAM

13件のポスト

• Fuzzy Metaballs+ use 3D Gaussians for differentiable rendering, integrating optical flow and producing watertight meshes. These techniques ensure spee… (♥ 185)

S-NeRF is a system that uses NeRFs to synthesize expansive and realistic street views for applications like autonomous driving simulation, robotics, a... (♥ 111)

LiDAR SLAM finds local geometric data important, but its computation is time-intensive. The Ring FALS method uses ring structure for swift normal esti... (♥ 53)

NeO 360 uses a hybrid representation to efficiently create 360° outdoor scenes from few RGB images, merging voxel and bird's-eye views for broad gener... (♥ 41)

DeepIPCv2 utilizes LiDAR point clouds for autonomous driving, offering consistent perception in low-illumination. Unaffected by lighting changes, it e... (♥ 38)

🚗 自動運転

1件のポスト

AnomalyGPT is an advanced Industrial Anomaly Detection method combining a LVLM and image encoder. It detects anomalies in industrial images without ma... (♥ 10)

🤖 ロボティクス

2件のポスト

DEnKF(Differentiable Ensemble Kalman Filters) uses stochastic neural networks for robot state estimation, modeling process noise. In real-world tasks ... (♥ 71)

DEnKF（Differentiable Ensemble Kalman Filters）は、ロボットの状態推定のために確率的ニューラルネットワークを使用し、プロセスノイズをモデル化する。ノイズが多い実世界のタスクで、位置誤差を59%削減。（IROS 2023） (♥ 44)

📄 論文紹介

2件のポスト

ENVIDR is introduced as a new framework that uses a decomposed neural renderer and an SDF-based surface model to accurately render and reconstruct sce... (♥ 42)

This study introduces a novel method for pose-graph optimization using a multiresolution representation built on a spanning tree. The approach allows ... (♥ 30)

💬 その他

2件のポスト

• Most Vision-and-Language Navigation methods use recurrent states, topological maps, or top-down semantic maps for memory. The new approach uses Grid M… (♥ 76)

Graduated Non-Convexity (GNC) に基づいたロバストなPose Graph Optimization。 Bスプラインを用いた形状関数を使用して、ロバストなカーネル形状を最適化することで、GNCの反復を削減し、精度を損なうことなく計算速度を向上させるている。 Adaptive... (♥ 4)

📋 全ポスト一覧

全20件を表示

| # | ポスト | RT | ♥ | 日付 | |--:|:-------|---:|--:|:-----| | 1 | [Fuzzy Metaballs+ use 3D Gaussians for differentiable rendering, integrating opti...](https://x.com/rsasaki0109/status/1697393915803181247) | 36 | 185 | 2023-08-31 | | 2 | [S-NeRF is a system that uses NeRFs to synthesize expansive and realistic street ...](https://x.com/rsasaki0109/status/1696489686339121280) | 23 | 111 | 2023-08-29 | | 3 | [Most Vision-and-Language Navigation methods use recurrent states, topological ma...](https://x.com/rsasaki0109/status/1694860738882662825) | 13 | 76 | 2023-08-24 | | 4 | [DEnKF(Differentiable Ensemble Kalman Filters) uses stochastic neural networks fo...](https://x.com/rsasaki0109/status/1694503194368639137) | 14 | 71 | 2023-08-24 | | 5 | [LiDAR SLAM finds local geometric data important, but its computation is time-int...](https://x.com/rsasaki0109/status/1697175899849027958) | 11 | 53 | 2023-08-31 | | 6 | [DEnKF（Differentiable Ensemble Kalman Filters）は、ロボットの状態推定のために確率的ニューラルネットワークを使用し、プ...](https://x.com/rsasaki0109/status/1695067638496403692) | 6 | 44 | 2023-08-25 | | 7 | [ENVIDR is introduced as a new framework that uses a decomposed neural renderer a...](https://x.com/rsasaki0109/status/1696495761876504706) | 6 | 42 | 2023-08-29 | | 8 | [NeO 360 uses a hybrid representation to efficiently create 360° outdoor scenes f...](https://x.com/rsasaki0109/status/1694912269203980310) | 8 | 41 | 2023-08-25 | | 9 | [DeepIPCv2 utilizes LiDAR point clouds for autonomous driving, offering consisten...](https://x.com/rsasaki0109/status/1694554684273680431) | 7 | 38 | 2023-08-24 | | 10 | [3D Gaussiansを使ったDynamic View Synthesisと6-DOFトラッキング Dynamic 3D Gaussians: Trackin...](https://x.com/rsasaki0109/status/1695300755828379722) | 6 | 37 | 2023-08-26 | | 11 | [This study introduces a novel method for pose-graph optimization using a multire...](https://x.com/rsasaki0109/status/1695230475122741415) | 4 | 30 | 2023-08-26 | | 12 | [Joint Intrinsic and Extrinsic LiDAR-Camera Calibration in Targetless Environment...](https://x.com/rsasaki0109/status/1696487625916588470) | 7 | 29 | 2023-08-29 | | 13 | [plants_temporal_matcher Estimating 4D Data Associations Towards Spatial-Temporal...](https://x.com/rsasaki0109/status/1697172658037936151) | 5 | 24 | 2023-08-31 | | 14 | [An proposed image-matching method uses topic-modeling for challenging scenes and...](https://x.com/rsasaki0109/status/1695069861381022073) | 2 | 23 | 2023-08-25 | | 15 | [高度なoccupancy mappingフレームワークであるWavemapは、ハールウェーブレット圧縮と階層的に計測値を統合することで、高い計算効率を実現し、細...](https://x.com/rsasaki0109/status/1696083597068001335) | 3 | 22 | 2023-08-28 | | 16 | [Robust Incremental Smoothing and Mapping(riSAM)は外れ値や誤った計測値を処理するためにGraduated Non-...](https://x.com/rsasaki0109/status/1695244285019656408) | 4 | 20 | 2023-08-26 | | 17 | [A region-wise hash map structure (RH-Map) is introduced, offering mapping and th...](https://x.com/rsasaki0109/status/1696498658785820672) | 2 | 19 | 2023-08-29 | | 18 | [バンドル調整をゲームとして捉え、強化学習を使用してダンピング係数を選択し、必要な反復回数を大幅に削減した手法 A Game of Bundle Adjustme...](https://x.com/rsasaki0109/status/1697386132307341498) | 4 | 14 | 2023-08-31 | | 19 | [AnomalyGPT is an advanced Industrial Anomaly Detection method combining a LVLM a...](https://x.com/rsasaki0109/status/1697205183212834847) | 2 | 10 | 2023-08-31 | | 20 | [Graduated Non-Convexity (GNC) に基づいたロバストなPose Graph Optimization。 Bスプラインを用いた形状関数を...](https://x.com/rsasaki0109/status/1695227626502459673) | 3 | 4 | 2023-08-26 |