Beyond Confidence: Exploiting Homogeneous Pattern for Semi-Supervised Semantic Segmentation
Authors: Rui Sun, Huayu Mai, Wangkai Li, Yujia Chen, Naisong Luo, Yuan Wang, Tianzhu Zhang
ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results on challenging benchmarks show that integrating Ag Score into existing semi-supervised segmentation frameworks yields consistent improvements. Also, sections like '5. Experiments' and tables (Table 1, Table 2, Table 3, Table 4) detailing quantitative results on various datasets confirm empirical studies. |
| Researcher Affiliation | Academia | 1Shenzhen International Graduate School, Tsinghua University 2Mo E Key Laboratory of Brain-inspired Intelligent Perception and Cognition, University of Science and Technology of China 3National Key Laboratory of Deep Space Exploration, Deep Space Exploration Laboratory. Correspondence to: Tianzhu Zhang <EMAIL>. All listed affiliations are academic institutions or affiliated public research labs, and the email domain `ustc.edu.cn` is academic. |
| Pseudocode | Yes | Algorithm 1 Agent Construction. Inputs: Pixel Embeddings F RB HW C, Prediction P RB HW , #Positive Agents N, #Negative Agents M Output: Positive Agents Ap RN C, Negative Agents An RM C and Algorithm 2 Orthogonal Selection Strategy. Inputs: Pixel Embedding F RM C, #Agents N Output: Agents A RN C are provided in Appendix A. |
| Open Source Code | No | The paper does not contain an explicit statement about releasing the source code for the described methodology, nor does it provide a direct link to a code repository. Phrases like 'We release our code' or links to platforms like GitHub are absent. |
| Open Datasets | Yes | Datasets: (1) PASCAL VOC 2012 (Everingham et al., 2010) is an object-centric semantic segmentation dataset... (2) Cityscapes (Cordts et al., 2016) is an urban scene understanding dataset... (3) COCO (Lin et al., 2014), composed of 118k/5k training/validation images... |
| Dataset Splits | Yes | We report m Io U (%) under various partition protocols... 1/16 (92) 1/8 (183) 1/4 (366) 1/2 (732) Full (1464) (Table 1). Similar partition protocols are mentioned for other datasets in Tables 2, 3, and 4. |
| Hardware Specification | Yes | The model is trained for 80 epochs on PASCAL and 240 epochs on Cityscapes with a batch size of 8, using 8 RTX 3090 GPUs (memory is 24G/GPU). |
| Software Dependencies | No | The paper mentions using Res Net-50/101 as the backbone and Deep Labv3+ as the decoder, and stochastic gradient descent (SGD) optimizer. However, it does not provide specific version numbers for any software, libraries, or frameworks used, which are required for reproducibility. |
| Experiment Setup | Yes | The crop size is set as 513 × 513 for PASCAL and 801 × 801 for Cityscapes, respectively. We adopt stochastic gradient descent (SGD) optimizer with an initial learning rate of 0.001 for PASCAL and 0.005 for Cityscapes. Polynomial Decay learning rate policy is applied throughout the whole training... We set the number of positive agents N = 64 and the number of negative agents M = 256 for all experiments. The model is trained for 80 epochs on PASCAL and 240 epochs on Cityscapes with a batch size of 8... |