Diffuse&Refine: Intrinsic Knowledge Generation and Aggregation for Incremental Object Detection
Authors: Jianzhou Wang, Yirui Wu, Lixin Yuan, Wenxiao Zhang, Jun Liu, Junyang Chen, Huan Wang, Wenhai Wang
IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on MS COCO dataset show Diff KA achieves state-of-the-art performance on IOD tasks with significant advantages. |
| Researcher Affiliation | Academia | 1College of Computer Science and Software Engineering, Hohai University 2Key Laboratory of Water Big Data Technology of Ministry of Water Resources, Hohai University 3School of Computing and Communication, Lancaster University 4College of Computer Science and Software Engineering, Shenzhen University 5College of Informatics, Huazhong Agricultural University 6Multimedia Laboratory, The Chinese University of Hong Kong EMAIL, EMAIL, EMAIL, EMAIL, EMAIL, EMAIL |
| Pseudocode | No | The paper describes methods in sections like '3 Methodology' and sub-sections such as '3.3 Intrinsic Knowledge Diffusion Module', which explain processes using descriptive text and mathematical equations (e.g., Eq. 1, 2, 3, 4, 6, 7), but it does not present any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing the source code for the Diff KA methodology, nor does it provide a link to a code repository. |
| Open Datasets | Yes | We evaluate Diff KA with the widely-used COCO 2017 dataset [Lin et al., 2014]. |
| Dataset Splits | Yes | In two-phase setting, the model is trained on A class using A A+B training samples, and the remaining B A+B samples are used to train the model on the new B class during the incremental learning. In experiments, we define two-phase settings, i.e., A = 40, B = 40, and A = 70, B = 10, resulting in total training phases M = 2, the number of classes in the first learning phase |C1| = A, and the number of classes in the second learning phase |C2| = B. In multi-phase setting, the model requires to recognize P + X Y classes, where the model is trained with P classes in the first learning phase, and then incrementally learns X new classes in each learning phase. Therefore, we could define M = Y + 1, |C1| = P, and |C2| = = |CM| = X. In our experiments, we set 40 + 20 2 and 40 + 10 4 for multi-phase experiments. |
| Hardware Specification | No | The paper mentions support from 'High Performance Computing Platform, Hohai University' in the acknowledgements, but it does not provide specific hardware details such as exact GPU/CPU models or memory specifications used for running experiments. |
| Software Dependencies | No | The paper states that 'Diff KA is bulit based on Deformable-DETR [Zhu et al., 2021] with its original settings', but it does not specify any particular software libraries or frameworks with their version numbers (e.g., Python, PyTorch, CUDA versions) that would be needed to replicate the experiment. |
| Experiment Setup | Yes | In incremental phases, we freeze the coarse Deformable-DETR and initialize a new detector with its parameters to train for 50 epochs in each phase. Table 4 presents results of Diff KA with different parameters η and λ1, λ2. η = 0.9 ensures the best performance of Diff KA... Meanwhile, Diff KA is not sensitive to the varying of λ1 and λ2 with high robustness. |