Publications
Search
1.
Xiong, Xinmiao; Liu, Bangya; Wang, Hao; Li, Dayou; Chen, Nuo; Feng, Andrew; Ding, Mingyu; Banerjee, Suman; Zhou, Yang; Fan, Zhiwen
Accelerating Transformer-Based Monocular SLAM via Geometric Utility Scoring Miscellaneous
2026, (arXiv:2604.08718 [cs.CV]).
@misc{xiong_accelerating_2026,
title = {Accelerating Transformer-Based Monocular SLAM via Geometric Utility Scoring},
author = {Xinmiao Xiong and Bangya Liu and Hao Wang and Dayou Li and Nuo Chen and Andrew Feng and Mingyu Ding and Suman Banerjee and Yang Zhou and Zhiwen Fan},
url = {http://arxiv.org/abs/2604.08718},
doi = {10.48550/arXiv.2604.08718},
year = {2026},
date = {2026-04-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {Geometric Foundation Models (GFMs) have recently advanced monocular SLAM by providing robust, calibration-free 3D priors. However, deploying these models on dense video streams introduces significant computational redundancy. Current GFM-based SLAM systems typically rely on post hoc keyframe selection. Because of this, they must perform expensive dense geometric decoding simply to determine whether a frame contains novel geometry, resulting in late rejection and wasted computation. To mitigate this inefficiency, we propose LeanGate, a lightweight feed-forward frame-gating network. LeanGate predicts a geometric utility score to assess a frame's mapping value prior to the heavy GFM feature extraction and matching stages. As a predictive plug-and-play module, our approach bypasses over 90% of redundant frames. Evaluations on standard SLAM benchmarks demonstrate that LeanGate reduces tracking FLOPs by more than 85% and achieves a 5x end-to-end throughput speedup. Furthermore, it maintains the tracking and mapping accuracy of dense baselines. Project page: https://lean-gate.github.io/},
note = {arXiv:2604.08718 [cs.CV]},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Geometric Foundation Models (GFMs) have recently advanced monocular SLAM by providing robust, calibration-free 3D priors. However, deploying these models on dense video streams introduces significant computational redundancy. Current GFM-based SLAM systems typically rely on post hoc keyframe selection. Because of this, they must perform expensive dense geometric decoding simply to determine whether a frame contains novel geometry, resulting in late rejection and wasted computation. To mitigate this inefficiency, we propose LeanGate, a lightweight feed-forward frame-gating network. LeanGate predicts a geometric utility score to assess a frame's mapping value prior to the heavy GFM feature extraction and matching stages. As a predictive plug-and-play module, our approach bypasses over 90% of redundant frames. Evaluations on standard SLAM benchmarks demonstrate that LeanGate reduces tracking FLOPs by more than 85% and achieves a 5x end-to-end throughput speedup. Furthermore, it maintains the tracking and mapping accuracy of dense baselines. Project page: https://lean-gate.github.io/
2.
Xiong, Butian; Liu, Rong; Zhou, Tiantian; Chen, Meida; Fan, Zhiwen; Feng, Andrew
NanoGS: Training-Free Gaussian Splat Simplification Miscellaneous
2026, (arXiv:2603.16103 [cs.CV]).
@misc{xiong_nanogs_2026,
title = {NanoGS: Training-Free Gaussian Splat Simplification},
author = {Butian Xiong and Rong Liu and Tiantian Zhou and Meida Chen and Zhiwen Fan and Andrew Feng},
url = {http://arxiv.org/abs/2603.16103},
doi = {10.48550/arXiv.2603.16103},
year = {2026},
date = {2026-03-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {3D Gaussian Splat (3DGS) enables high-fidelity, real-time novel view synthesis by representing scenes with large sets of anisotropic primitives, but often requires millions of Splats, incurring significant storage and transmission costs. Most existing compression methods rely on GPU-intensive post-training optimization with calibrated images, limiting practical deployment. We introduce NanoGS, a training-free and lightweight framework for Gaussian Splat simplification. Instead of relying on image-based rendering supervision, NanoGS formulates simplification as local pairwise merging over a sparse spatial graph. The method approximates a pair of Gaussians with a single primitive using mass preserved moment matching and evaluates merge quality through a principled merge cost between the original mixture and its approximation. By restricting merge candidates to local neighborhoods and selecting compatible pairs efficiently, NanoGS produces compact Gaussian representations while preserving scene structure and appearance. NanoGS operates directly on existing Gaussian Splat models, runs efficiently on CPU, and preserves the standard 3DGS parameterization, enabling seamless integration with existing rendering pipelines. Experiments demonstrate that NanoGS substantially reduces primitive count while maintaining high rendering fidelity, providing an efficient and practical solution for Gaussian Splat simplification. Our project website is available at https://saliteta.github.io/NanoGS/.},
note = {arXiv:2603.16103 [cs.CV]},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
3D Gaussian Splat (3DGS) enables high-fidelity, real-time novel view synthesis by representing scenes with large sets of anisotropic primitives, but often requires millions of Splats, incurring significant storage and transmission costs. Most existing compression methods rely on GPU-intensive post-training optimization with calibrated images, limiting practical deployment. We introduce NanoGS, a training-free and lightweight framework for Gaussian Splat simplification. Instead of relying on image-based rendering supervision, NanoGS formulates simplification as local pairwise merging over a sparse spatial graph. The method approximates a pair of Gaussians with a single primitive using mass preserved moment matching and evaluates merge quality through a principled merge cost between the original mixture and its approximation. By restricting merge candidates to local neighborhoods and selecting compatible pairs efficiently, NanoGS produces compact Gaussian representations while preserving scene structure and appearance. NanoGS operates directly on existing Gaussian Splat models, runs efficiently on CPU, and preserves the standard 3DGS parameterization, enabling seamless integration with existing rendering pipelines. Experiments demonstrate that NanoGS substantially reduces primitive count while maintaining high rendering fidelity, providing an efficient and practical solution for Gaussian Splat simplification. Our project website is available at https://saliteta.github.io/NanoGS/.
3.
Chen, Pengcheng; Hu, Yue; Li, Wenhao; Gunderson, Nicole M.; Feng, Andrew; Sun, Zhenglong; Beerel, Peter; Seibel, Eric J.
NeVStereo: A NeRF-Driven NVS-Stereo Architecture for High-Fidelity 3D Tasks Miscellaneous
2026, (arXiv:2602.05423 [cs.CV]).
@misc{chen_nevstereo_2026,
title = {NeVStereo: A NeRF-Driven NVS-Stereo Architecture for High-Fidelity 3D Tasks},
author = {Pengcheng Chen and Yue Hu and Wenhao Li and Nicole M. Gunderson and Andrew Feng and Zhenglong Sun and Peter Beerel and Eric J. Seibel},
url = {http://arxiv.org/abs/2602.05423},
doi = {10.48550/arXiv.2602.05423},
year = {2026},
date = {2026-02-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {In modern dense 3D reconstruction, feed-forward systems (e.g., VGGT, pi3) focus on end-to-end matching and geometry prediction but do not explicitly output the novel view synthesis (NVS). Neural rendering-based approaches offer high-fidelity NVS and detailed geometry from posed images, yet they typically assume fixed camera poses and can be sensitive to pose errors. As a result, it remains non-trivial to obtain a single framework that can offer accurate poses, reliable depth, high-quality rendering, and accurate 3D surfaces from casually captured views. We present NeVStereo, a NeRF-driven NVS-stereo architecture that aims to jointly deliver camera poses, multi-view depth, novel view synthesis, and surface reconstruction from multi-view RGB-only inputs. NeVStereo combines NeRF-based NVS for stereo-friendly renderings, confidence-guided multi-view depth estimation, NeRF-coupled bundle adjustment for pose refinement, and an iterative refinement stage that updates both depth and the radiance field to improve geometric consistency. This design mitigated the common NeRF-based issues such as surface stacking, artifacts, and pose-depth coupling. Across indoor, outdoor, tabletop, and aerial benchmarks, our experiments indicate that NeVStereo achieves consistently strong zero-shot performance, with up to 36% lower depth error, 10.4% improved pose accuracy, 4.5% higher NVS fidelity, and state-of-the-art mesh quality (F1 91.93%, Chamfer 4.35 mm) compared to existing prestigious methods.},
note = {arXiv:2602.05423 [cs.CV]},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
In modern dense 3D reconstruction, feed-forward systems (e.g., VGGT, pi3) focus on end-to-end matching and geometry prediction but do not explicitly output the novel view synthesis (NVS). Neural rendering-based approaches offer high-fidelity NVS and detailed geometry from posed images, yet they typically assume fixed camera poses and can be sensitive to pose errors. As a result, it remains non-trivial to obtain a single framework that can offer accurate poses, reliable depth, high-quality rendering, and accurate 3D surfaces from casually captured views. We present NeVStereo, a NeRF-driven NVS-stereo architecture that aims to jointly deliver camera poses, multi-view depth, novel view synthesis, and surface reconstruction from multi-view RGB-only inputs. NeVStereo combines NeRF-based NVS for stereo-friendly renderings, confidence-guided multi-view depth estimation, NeRF-coupled bundle adjustment for pose refinement, and an iterative refinement stage that updates both depth and the radiance field to improve geometric consistency. This design mitigated the common NeRF-based issues such as surface stacking, artifacts, and pose-depth coupling. Across indoor, outdoor, tabletop, and aerial benchmarks, our experiments indicate that NeVStereo achieves consistently strong zero-shot performance, with up to 36% lower depth error, 10.4% improved pose accuracy, 4.5% higher NVS fidelity, and state-of-the-art mesh quality (F1 91.93%, Chamfer 4.35 mm) compared to existing prestigious methods.
4.
Jin, Zhangyu; Feng, Andrew; Chemburkar, Ankur; Melo, Celso M. De
PromptGAR: Flexible Promptive Group Activity Recognition Miscellaneous
2025, (arXiv:2503.08933 [cs.CV]).
@misc{jin_promptgar_2025-1,
title = {PromptGAR: Flexible Promptive Group Activity Recognition},
author = {Zhangyu Jin and Andrew Feng and Ankur Chemburkar and Celso M. De Melo},
url = {http://arxiv.org/abs/2503.08933},
doi = {10.48550/arXiv.2503.08933},
year = {2025},
date = {2025-08-01},
urldate = {2026-06-19},
publisher = {arXiv},
abstract = {We present PromptGAR, a novel framework for Group Activity Recognition (GAR) that offering both input flexibility and high recognition accuracy. The existing approaches suffer from limited real-world applicability due to their reliance on full prompt annotations, fixed number of frames and instances, and the lack of actor consistency. To bridge the gap, we proposed PromptGAR, which is the first GAR model to provide input flexibility across prompts, frames, and instances without the need for retraining. We leverage diverse visual prompts, like bounding boxes, skeletal keypoints, and instance identities, by unifying them as point prompts. A recognition decoder then cross-updates class and prompt tokens for enhanced performance. To ensure actor consistency for extended activity durations, we also introduce a relative instance attention mechanism that directly encodes instance identities. Comprehensive evaluations demonstrate that PromptGAR achieves competitive performances both on full prompts and partial prompt inputs, establishing its effectiveness on input flexibility and generalization ability for real-world applications.},
note = {arXiv:2503.08933 [cs.CV]},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
We present PromptGAR, a novel framework for Group Activity Recognition (GAR) that offering both input flexibility and high recognition accuracy. The existing approaches suffer from limited real-world applicability due to their reliance on full prompt annotations, fixed number of frames and instances, and the lack of actor consistency. To bridge the gap, we proposed PromptGAR, which is the first GAR model to provide input flexibility across prompts, frames, and instances without the need for retraining. We leverage diverse visual prompts, like bounding boxes, skeletal keypoints, and instance identities, by unifying them as point prompts. A recognition decoder then cross-updates class and prompt tokens for enhanced performance. To ensure actor consistency for extended activity durations, we also introduce a relative instance attention mechanism that directly encodes instance identities. Comprehensive evaluations demonstrate that PromptGAR achieves competitive performances both on full prompts and partial prompt inputs, establishing its effectiveness on input flexibility and generalization ability for real-world applications.
5.
Jin, Zhangyu; Feng, Andrew
SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors Miscellaneous
2025, (arXiv:2508.18531 [cs.CV]).
@misc{jin_sat-skylines_2025,
title = {SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors},
author = {Zhangyu Jin and Andrew Feng},
url = {http://arxiv.org/abs/2508.18531},
doi = {10.48550/arXiv.2508.18531},
year = {2025},
date = {2025-08-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {We present SatSkylines, a 3D building generation approach that takes satellite imagery and coarse geometric priors. Without proper geometric guidance, existing image-based 3D generation methods struggle to recover accurate building structures from the top-down views of satellite images alone. On the other hand, 3D detailization methods tend to rely heavily on highly detailed voxel inputs and fail to produce satisfying results from simple priors such as cuboids. To address these issues, our key idea is to model the transformation from interpolated noisy coarse priors to detailed geometries, enabling flexible geometric control without additional computational cost. We have further developed Skylines-50K, a large-scale dataset of over 50,000 unique and stylized 3D building assets in order to support the generations of detailed building models. Extensive evaluations indicate the effectiveness of our model and strong generalization ability.},
note = {arXiv:2508.18531 [cs.CV]},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
We present SatSkylines, a 3D building generation approach that takes satellite imagery and coarse geometric priors. Without proper geometric guidance, existing image-based 3D generation methods struggle to recover accurate building structures from the top-down views of satellite images alone. On the other hand, 3D detailization methods tend to rely heavily on highly detailed voxel inputs and fail to produce satisfying results from simple priors such as cuboids. To address these issues, our key idea is to model the transformation from interpolated noisy coarse priors to detailed geometries, enabling flexible geometric control without additional computational cost. We have further developed Skylines-50K, a large-scale dataset of over 50,000 unique and stylized 3D building assets in order to support the generations of detailed building models. Extensive evaluations indicate the effectiveness of our model and strong generalization ability.
Filter
2026
Xiong, Xinmiao; Liu, Bangya; Wang, Hao; Li, Dayou; Chen, Nuo; Feng, Andrew; Ding, Mingyu; Banerjee, Suman; Zhou, Yang; Fan, Zhiwen
Accelerating Transformer-Based Monocular SLAM via Geometric Utility Scoring Miscellaneous
2026, (arXiv:2604.08718 [cs.CV]).
Abstract | Links | BibTeX | Tags: AR, Geospatial
@misc{xiong_accelerating_2026,
title = {Accelerating Transformer-Based Monocular SLAM via Geometric Utility Scoring},
author = {Xinmiao Xiong and Bangya Liu and Hao Wang and Dayou Li and Nuo Chen and Andrew Feng and Mingyu Ding and Suman Banerjee and Yang Zhou and Zhiwen Fan},
url = {http://arxiv.org/abs/2604.08718},
doi = {10.48550/arXiv.2604.08718},
year = {2026},
date = {2026-04-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {Geometric Foundation Models (GFMs) have recently advanced monocular SLAM by providing robust, calibration-free 3D priors. However, deploying these models on dense video streams introduces significant computational redundancy. Current GFM-based SLAM systems typically rely on post hoc keyframe selection. Because of this, they must perform expensive dense geometric decoding simply to determine whether a frame contains novel geometry, resulting in late rejection and wasted computation. To mitigate this inefficiency, we propose LeanGate, a lightweight feed-forward frame-gating network. LeanGate predicts a geometric utility score to assess a frame's mapping value prior to the heavy GFM feature extraction and matching stages. As a predictive plug-and-play module, our approach bypasses over 90% of redundant frames. Evaluations on standard SLAM benchmarks demonstrate that LeanGate reduces tracking FLOPs by more than 85% and achieves a 5x end-to-end throughput speedup. Furthermore, it maintains the tracking and mapping accuracy of dense baselines. Project page: https://lean-gate.github.io/},
note = {arXiv:2604.08718 [cs.CV]},
keywords = {AR, Geospatial},
pubstate = {published},
tppubtype = {misc}
}
Geometric Foundation Models (GFMs) have recently advanced monocular SLAM by providing robust, calibration-free 3D priors. However, deploying these models on dense video streams introduces significant computational redundancy. Current GFM-based SLAM systems typically rely on post hoc keyframe selection. Because of this, they must perform expensive dense geometric decoding simply to determine whether a frame contains novel geometry, resulting in late rejection and wasted computation. To mitigate this inefficiency, we propose LeanGate, a lightweight feed-forward frame-gating network. LeanGate predicts a geometric utility score to assess a frame's mapping value prior to the heavy GFM feature extraction and matching stages. As a predictive plug-and-play module, our approach bypasses over 90% of redundant frames. Evaluations on standard SLAM benchmarks demonstrate that LeanGate reduces tracking FLOPs by more than 85% and achieves a 5x end-to-end throughput speedup. Furthermore, it maintains the tracking and mapping accuracy of dense baselines. Project page: https://lean-gate.github.io/
Xiong, Butian; Liu, Rong; Zhou, Tiantian; Chen, Meida; Fan, Zhiwen; Feng, Andrew
NanoGS: Training-Free Gaussian Splat Simplification Miscellaneous
2026, (arXiv:2603.16103 [cs.CV]).
Abstract | Links | BibTeX | Tags: Geospatial, Graphics
@misc{xiong_nanogs_2026,
title = {NanoGS: Training-Free Gaussian Splat Simplification},
author = {Butian Xiong and Rong Liu and Tiantian Zhou and Meida Chen and Zhiwen Fan and Andrew Feng},
url = {http://arxiv.org/abs/2603.16103},
doi = {10.48550/arXiv.2603.16103},
year = {2026},
date = {2026-03-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {3D Gaussian Splat (3DGS) enables high-fidelity, real-time novel view synthesis by representing scenes with large sets of anisotropic primitives, but often requires millions of Splats, incurring significant storage and transmission costs. Most existing compression methods rely on GPU-intensive post-training optimization with calibrated images, limiting practical deployment. We introduce NanoGS, a training-free and lightweight framework for Gaussian Splat simplification. Instead of relying on image-based rendering supervision, NanoGS formulates simplification as local pairwise merging over a sparse spatial graph. The method approximates a pair of Gaussians with a single primitive using mass preserved moment matching and evaluates merge quality through a principled merge cost between the original mixture and its approximation. By restricting merge candidates to local neighborhoods and selecting compatible pairs efficiently, NanoGS produces compact Gaussian representations while preserving scene structure and appearance. NanoGS operates directly on existing Gaussian Splat models, runs efficiently on CPU, and preserves the standard 3DGS parameterization, enabling seamless integration with existing rendering pipelines. Experiments demonstrate that NanoGS substantially reduces primitive count while maintaining high rendering fidelity, providing an efficient and practical solution for Gaussian Splat simplification. Our project website is available at https://saliteta.github.io/NanoGS/.},
note = {arXiv:2603.16103 [cs.CV]},
keywords = {Geospatial, Graphics},
pubstate = {published},
tppubtype = {misc}
}
3D Gaussian Splat (3DGS) enables high-fidelity, real-time novel view synthesis by representing scenes with large sets of anisotropic primitives, but often requires millions of Splats, incurring significant storage and transmission costs. Most existing compression methods rely on GPU-intensive post-training optimization with calibrated images, limiting practical deployment. We introduce NanoGS, a training-free and lightweight framework for Gaussian Splat simplification. Instead of relying on image-based rendering supervision, NanoGS formulates simplification as local pairwise merging over a sparse spatial graph. The method approximates a pair of Gaussians with a single primitive using mass preserved moment matching and evaluates merge quality through a principled merge cost between the original mixture and its approximation. By restricting merge candidates to local neighborhoods and selecting compatible pairs efficiently, NanoGS produces compact Gaussian representations while preserving scene structure and appearance. NanoGS operates directly on existing Gaussian Splat models, runs efficiently on CPU, and preserves the standard 3DGS parameterization, enabling seamless integration with existing rendering pipelines. Experiments demonstrate that NanoGS substantially reduces primitive count while maintaining high rendering fidelity, providing an efficient and practical solution for Gaussian Splat simplification. Our project website is available at https://saliteta.github.io/NanoGS/.
Chen, Pengcheng; Hu, Yue; Li, Wenhao; Gunderson, Nicole M.; Feng, Andrew; Sun, Zhenglong; Beerel, Peter; Seibel, Eric J.
NeVStereo: A NeRF-Driven NVS-Stereo Architecture for High-Fidelity 3D Tasks Miscellaneous
2026, (arXiv:2602.05423 [cs.CV]).
Abstract | Links | BibTeX | Tags: Geospatial, Graphics
@misc{chen_nevstereo_2026,
title = {NeVStereo: A NeRF-Driven NVS-Stereo Architecture for High-Fidelity 3D Tasks},
author = {Pengcheng Chen and Yue Hu and Wenhao Li and Nicole M. Gunderson and Andrew Feng and Zhenglong Sun and Peter Beerel and Eric J. Seibel},
url = {http://arxiv.org/abs/2602.05423},
doi = {10.48550/arXiv.2602.05423},
year = {2026},
date = {2026-02-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {In modern dense 3D reconstruction, feed-forward systems (e.g., VGGT, pi3) focus on end-to-end matching and geometry prediction but do not explicitly output the novel view synthesis (NVS). Neural rendering-based approaches offer high-fidelity NVS and detailed geometry from posed images, yet they typically assume fixed camera poses and can be sensitive to pose errors. As a result, it remains non-trivial to obtain a single framework that can offer accurate poses, reliable depth, high-quality rendering, and accurate 3D surfaces from casually captured views. We present NeVStereo, a NeRF-driven NVS-stereo architecture that aims to jointly deliver camera poses, multi-view depth, novel view synthesis, and surface reconstruction from multi-view RGB-only inputs. NeVStereo combines NeRF-based NVS for stereo-friendly renderings, confidence-guided multi-view depth estimation, NeRF-coupled bundle adjustment for pose refinement, and an iterative refinement stage that updates both depth and the radiance field to improve geometric consistency. This design mitigated the common NeRF-based issues such as surface stacking, artifacts, and pose-depth coupling. Across indoor, outdoor, tabletop, and aerial benchmarks, our experiments indicate that NeVStereo achieves consistently strong zero-shot performance, with up to 36% lower depth error, 10.4% improved pose accuracy, 4.5% higher NVS fidelity, and state-of-the-art mesh quality (F1 91.93%, Chamfer 4.35 mm) compared to existing prestigious methods.},
note = {arXiv:2602.05423 [cs.CV]},
keywords = {Geospatial, Graphics},
pubstate = {published},
tppubtype = {misc}
}
In modern dense 3D reconstruction, feed-forward systems (e.g., VGGT, pi3) focus on end-to-end matching and geometry prediction but do not explicitly output the novel view synthesis (NVS). Neural rendering-based approaches offer high-fidelity NVS and detailed geometry from posed images, yet they typically assume fixed camera poses and can be sensitive to pose errors. As a result, it remains non-trivial to obtain a single framework that can offer accurate poses, reliable depth, high-quality rendering, and accurate 3D surfaces from casually captured views. We present NeVStereo, a NeRF-driven NVS-stereo architecture that aims to jointly deliver camera poses, multi-view depth, novel view synthesis, and surface reconstruction from multi-view RGB-only inputs. NeVStereo combines NeRF-based NVS for stereo-friendly renderings, confidence-guided multi-view depth estimation, NeRF-coupled bundle adjustment for pose refinement, and an iterative refinement stage that updates both depth and the radiance field to improve geometric consistency. This design mitigated the common NeRF-based issues such as surface stacking, artifacts, and pose-depth coupling. Across indoor, outdoor, tabletop, and aerial benchmarks, our experiments indicate that NeVStereo achieves consistently strong zero-shot performance, with up to 36% lower depth error, 10.4% improved pose accuracy, 4.5% higher NVS fidelity, and state-of-the-art mesh quality (F1 91.93%, Chamfer 4.35 mm) compared to existing prestigious methods.
2025
Jin, Zhangyu; Feng, Andrew; Chemburkar, Ankur; Melo, Celso M. De
PromptGAR: Flexible Promptive Group Activity Recognition Miscellaneous
2025, (arXiv:2503.08933 [cs.CV]).
Abstract | Links | BibTeX | Tags: Geospatial
@misc{jin_promptgar_2025-1,
title = {PromptGAR: Flexible Promptive Group Activity Recognition},
author = {Zhangyu Jin and Andrew Feng and Ankur Chemburkar and Celso M. De Melo},
url = {http://arxiv.org/abs/2503.08933},
doi = {10.48550/arXiv.2503.08933},
year = {2025},
date = {2025-08-01},
urldate = {2026-06-19},
publisher = {arXiv},
abstract = {We present PromptGAR, a novel framework for Group Activity Recognition (GAR) that offering both input flexibility and high recognition accuracy. The existing approaches suffer from limited real-world applicability due to their reliance on full prompt annotations, fixed number of frames and instances, and the lack of actor consistency. To bridge the gap, we proposed PromptGAR, which is the first GAR model to provide input flexibility across prompts, frames, and instances without the need for retraining. We leverage diverse visual prompts, like bounding boxes, skeletal keypoints, and instance identities, by unifying them as point prompts. A recognition decoder then cross-updates class and prompt tokens for enhanced performance. To ensure actor consistency for extended activity durations, we also introduce a relative instance attention mechanism that directly encodes instance identities. Comprehensive evaluations demonstrate that PromptGAR achieves competitive performances both on full prompts and partial prompt inputs, establishing its effectiveness on input flexibility and generalization ability for real-world applications.},
note = {arXiv:2503.08933 [cs.CV]},
keywords = {Geospatial},
pubstate = {published},
tppubtype = {misc}
}
We present PromptGAR, a novel framework for Group Activity Recognition (GAR) that offering both input flexibility and high recognition accuracy. The existing approaches suffer from limited real-world applicability due to their reliance on full prompt annotations, fixed number of frames and instances, and the lack of actor consistency. To bridge the gap, we proposed PromptGAR, which is the first GAR model to provide input flexibility across prompts, frames, and instances without the need for retraining. We leverage diverse visual prompts, like bounding boxes, skeletal keypoints, and instance identities, by unifying them as point prompts. A recognition decoder then cross-updates class and prompt tokens for enhanced performance. To ensure actor consistency for extended activity durations, we also introduce a relative instance attention mechanism that directly encodes instance identities. Comprehensive evaluations demonstrate that PromptGAR achieves competitive performances both on full prompts and partial prompt inputs, establishing its effectiveness on input flexibility and generalization ability for real-world applications.
Jin, Zhangyu; Feng, Andrew
SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors Miscellaneous
2025, (arXiv:2508.18531 [cs.CV]).
Abstract | Links | BibTeX | Tags: Geospatial
@misc{jin_sat-skylines_2025,
title = {SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors},
author = {Zhangyu Jin and Andrew Feng},
url = {http://arxiv.org/abs/2508.18531},
doi = {10.48550/arXiv.2508.18531},
year = {2025},
date = {2025-08-01},
urldate = {2026-06-18},
publisher = {arXiv},
abstract = {We present SatSkylines, a 3D building generation approach that takes satellite imagery and coarse geometric priors. Without proper geometric guidance, existing image-based 3D generation methods struggle to recover accurate building structures from the top-down views of satellite images alone. On the other hand, 3D detailization methods tend to rely heavily on highly detailed voxel inputs and fail to produce satisfying results from simple priors such as cuboids. To address these issues, our key idea is to model the transformation from interpolated noisy coarse priors to detailed geometries, enabling flexible geometric control without additional computational cost. We have further developed Skylines-50K, a large-scale dataset of over 50,000 unique and stylized 3D building assets in order to support the generations of detailed building models. Extensive evaluations indicate the effectiveness of our model and strong generalization ability.},
note = {arXiv:2508.18531 [cs.CV]},
keywords = {Geospatial},
pubstate = {published},
tppubtype = {misc}
}
We present SatSkylines, a 3D building generation approach that takes satellite imagery and coarse geometric priors. Without proper geometric guidance, existing image-based 3D generation methods struggle to recover accurate building structures from the top-down views of satellite images alone. On the other hand, 3D detailization methods tend to rely heavily on highly detailed voxel inputs and fail to produce satisfying results from simple priors such as cuboids. To address these issues, our key idea is to model the transformation from interpolated noisy coarse priors to detailed geometries, enabling flexible geometric control without additional computational cost. We have further developed Skylines-50K, a large-scale dataset of over 50,000 unique and stylized 3D building assets in order to support the generations of detailed building models. Extensive evaluations indicate the effectiveness of our model and strong generalization ability.