Causal Intervention for Weakly-Supervised Semantic Segmentation
Authors: Dong Zhang, Hanwang Zhang, Jinhui Tang, Xian-Sheng Hua, Qianru Sun
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | On PASCAL VOC 2012 and MS-COCO, we show that CONTA boosts various popular WSSS methods to new state-of-the-arts. |
| Researcher Affiliation | Collaboration | 1School of Computer Science and Engineering, Nanjing University of Science and Technology; 2Nanyang Technological University; 3Damo Academy, Alibaba Group; 4Singapore Management University. |
| Pseudocode | No | The paper describes the steps of its proposed method (CONTA) in text, such as 'Step 1. Image Classification.', but it does not present any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | 1Code is open-sourced at: https://github.com/ZHANGDONG-NJUST/CONTA |
| Open Datasets | Yes | PASCAL VOC 2012 [12] contains 21 classes (one background class) which includes 1,464, 1,449 and 1,456 images for training, validation (val) and test, respectively. As the common practice in [1, 57], in our experiments, we used an enlarged training set with 10,582 images, where the extra images and labels are from [17]. MS-COCO [30] contains 81 classes (one background class), 80k, and 40k images for training and val. |
| Dataset Splits | Yes | PASCAL VOC 2012 [12] contains 21 classes (one background class) which includes 1,464, 1,449 and 1,456 images for training, validation (val) and test, respectively. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running experiments, such as GPU models, CPU types, or memory specifications. |
| Software Dependencies | No | The paper refers to various software components and models like Deep Lab-v2, SEAM, IRNet, DSRG, and SEC, and its code is open-sourced on GitHub (implying Python/PyTorch), but it does not specify exact version numbers for any key software dependencies or libraries. |
| Experiment Setup | No | The paper describes the iterative procedure of CONTA and its internal steps, including the number of rounds ('#round = 3') and where feature maps were concatenated, but it defers to 'the same settings as reported in the official codes' for baselines and does not explicitly list specific hyperparameters (e.g., learning rate, batch size, optimizer) used for training. |