SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2409.05162 · ECCV · 2024

Can OOD Object Detectors Learn from Foundation Models?

Xiaojuan Qi, Xin Wen, Shizhen Zhao, Yingxian Chen, Jiahui Liu

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 9 functions out of this paper's own repositories and ran 9 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
CVMI-Lab/SyncOOD canonical 9 of 9
FunctionStatusWhere it lives
add_args Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/interface.py
code served (permissive licence) · get_code("aec928841c66e96d")
adjust_block_compatibility Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/modeling/regnet_model.py
code served (permissive licence) · get_code("00564fe706e02386")
clamp_log_variance Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/modeling/modeling_utils.py
code served (permissive licence) · get_code("d2ac285b09b448cb")
conv2d Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/modeling/regnet_model.py
code served (permissive licence) · get_code("086b324b9fee208c")
covariance_output_to_cholesky Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/modeling/modeling_utils.py
code served (permissive licence) · get_code("5f5b17f6d41bc1b0")
get_probabilistic_loss_weight Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/modeling/modeling_utils.py
code served (permissive licence) · get_code("3aa86d0a61778718")
permute_to_N_HWA_K Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/inference/retinanet_predictor.py
code served (permissive licence) · get_code("573ab018aa3f2e6f")
pool2d Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/modeling/regnet_model.py
code served (permissive licence) · get_code("f0af8fecb0c26921")
set_model_defaults Ran CVMI-Lab/SyncOOD/OOD_OBJ_DET/core/detr_args.py
code served (permissive licence) · get_code("2b0907d9ff39c93b")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Out-of-distribution (OOD) object detection is a challenging task due to the absence of open-set OOD data. Inspired by recent advancements in text-to-image generative models, such as Stable Diffusion, we study the potential of generative models trained on large-scale openset data to synthesize OOD samples, thereby enhancing OOD object detection. We introduce SyncOOD, a simple data curation method that capitalizes on the capabilities of large foundation models to automatically extract meaningful OOD data from text-to-image generative models. This offers the model access to open-world knowledge encapsulated within off-the-shelf foundation models. The synthetic OOD samples are then employed to augment the training of a lightweight, plug-and-play OOD detector, thus effectively optimizing the in-distribution (ID)/OOD decision boundaries. Extensive experiments across multiple benchmarks demonstrate that SyncOOD significantly outperforms existing methods, establishing new state-of-the-art performance with minimal synthetic data usage. The project is available at https://github.com/CVMI-Lab/SyncOOD.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2409.05162")
get_code_for_paper("2409.05162")
have("2409.05162")

Connect an agent — have() is free.