Target-Free Text-Guided Image Manipulation

Authors: Wan-Cyuan Fan, Cheng-Fu Yang, Chiao-An Yang, Yu-Chiang Frank Wang

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct extensive experiments on the datasets of CLEVR and COCO, and the effectiveness and generalizability of our proposed method can be successfully verified.
Researcher Affiliation Collaboration Wan-Cyuan Fan1, Cheng-Fu Yang2, Chiao-An Yang3, Yu-Chiang Frank Wang1, 4 1National Taiwan University 2UCLA 3Purdue University 4NVIDIA
Pseudocode Yes For complete learning details (including pseudo code) of our c Mani GAN, please refer to the supplementary material.
Open Source Code No The paper provides a 'Project page: sites.google.com/view/wancyuanfan/ projects/cmanigan' but this is not explicitly stated to be a source code repository.
Open Datasets Yes The CLEVR dataset (Johnson et al. 2017) is created for multimodal learning tasks... The COCO dataset (Lin et al. 2014) contains 118k real-world scene images for training... We will make the datasets publicly available for reproduction and comparison purposes.
Dataset Splits Yes CLEVR... with about 28.1K/4.6K paired synthesized images for training/validation. COCO... with about 12K/3K samples for training/validation.
Hardware Specification No The paper mentions support from 'National Center for High-performance Computing (NCHC) and Taiwan Computing Cloud (TWCC) for providing computational and storage resources' but does not specify exact hardware models (e.g., GPU/CPU types).
Software Dependencies No The paper mentions software tools like 'Core NLP', 'BERT', 'T5', and 'NLTK tools' but does not provide specific version numbers for any of them.
Experiment Setup No The paper states 'Please refer to the supplementary materials for the full objectives and implementation details.' and 'For more ablation studies of objective function and reasoner module, please refer to our supplementary materials.', indicating that detailed experimental setup information is not present in the main text.