{"id":20377555,"url":"https://github.com/reshalfahsi/givted-net","last_synced_at":"2026-04-19T12:33:37.913Z","repository":{"id":244619388,"uuid":"815771412","full_name":"reshalfahsi/GIVTED-Net","owner":"reshalfahsi","description":"The official repository for \"GIVTED-Net: GhostNet-Mobile Involution ViT Encoder-Decoder Network for Lightweight Medical Image Segmentation.\"","archived":false,"fork":false,"pushed_at":"2024-06-16T18:37:55.000Z","size":200,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"master","last_synced_at":"2025-01-15T07:14:39.748Z","etag":null,"topics":["isic-2018","kvasir-instrument","lightweight-model","medical-image-segmentation","pytorch"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/reshalfahsi.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2024-06-16T05:20:54.000Z","updated_at":"2024-06-16T18:37:58.000Z","dependencies_parsed_at":"2025-01-15T06:37:21.041Z","dependency_job_id":"fe8cff5f-31ea-4f59-92d4-79069954290d","html_url":"https://github.com/reshalfahsi/GIVTED-Net","commit_stats":null,"previous_names":["reshalfahsi/givted-net"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reshalfahsi%2FGIVTED-Net","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reshalfahsi%2FGIVTED-Net/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reshalfahsi%2FGIVTED-Net/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reshalfahsi%2FGIVTED-Net/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/reshalfahsi","download_url":"https://codeload.github.com/reshalfahsi/GIVTED-Net/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":241928528,"owners_count":20043821,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["isic-2018","kvasir-instrument","lightweight-model","medical-image-segmentation","pytorch"],"created_at":"2024-11-15T01:45:36.949Z","updated_at":"2026-04-19T12:33:37.861Z","avatar_url":"https://github.com/reshalfahsi.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# GIVTED-Net\n\n\u003cdiv align=\"center\"\u003e\n\n\u003cimg src=\"https://ieeexplore.ieee.org/ielx8/6287639/10380310/10552270/graphical_abstract/access-gagraphic-3411870.jpg\" width=\"600\"\u003e\n\nThe architecture of GIVTED-Net.\n\u003c/div\u003e\n\nThis is the official repository for \"[GIVTED-Net: GhostNet-Mobile Involution ViT Encoder-Decoder Network for Lightweight Medical Image Segmentation](https://ieeexplore.ieee.org/document/10552270).\". Here, we introduce MIViT (Mobile Involution Vision Transformer) modules powered by newly formed transformer blocks, InvoFormer. The blocks follow the design principle of MetaFormer with an involution-based token mixer and a squeeze-and-excitation module-based channel MLP. These components, along with the Ghost bottlenecks of GhostNet, are compiled in the proposed medical image segmentation model, GIVTED-Net. Ghost bottlenecks and MIViT modules are employed in the encoder and decoder, respectively. The model is lightweight (0.19M of parameters and 0.56GFLOPs), fast (2.47 FPS on the Raspberry Pi 4), and quantitatively superior to prior models.\n\n\n## Requirements\n\nThis implementation requires some libraries and frameworks. They can easily be installed by:\n\n```bash\npip install -r requirements.txt\n```\n\n\n## Dataset\n\nWe use the ``KvasirInstrument``, ``ISIC2018``, and ``WBCImage`` datasets to assess the performance of the model. We arrange the datasets in the following:\n\n```bash\n.\n└── experiment/\n    ├── KvasirInstrument/\n    │   ├── TrainDataset/\n    │   │   ├── images/\n    │   │   │   ├── image_00.jpg\n    │   │   │   └── ...\n    │   │   └── masks/\n    │   │       ├── image_00.jpg\n    │   │       └── ...\n    │   ├── TestDataset/\n    │   │   ├── images/\n    │   │   │   ├── image_00.jpg\n    │   │   │   └── ...\n    │   │   └── masks/\n    │   │       ├── image_00.jpg\n    │   │       └── ...\n    │   ├── result/\n    │   │   ├── image_00.jpg\n    │   │   └── ...\n    │   └── model_pth/\n    │       ├── GIVTEDNet.pth\n    │       └── ...\n    ├── ISIC2018/\n    │   ├── TrainDataset/\n    │   │   ├── images/\n    │   │   │   ├── image_00.jpg\n    │   │   │   └── ...\n    │   │   └── masks/\n    │   │       ├── image_00.jpg\n    │   │       └── ...\n    │   ├── TestDataset/\n    │   │   ├── images/\n    │   │   │   ├── image_00.jpg\n    │   │   │   └── ...\n    │   │   └── masks/\n    │   │       ├── image_00.jpg\n    │   │       └── ...\n    │   ├── result/\n    │   │   ├── image_00.jpg\n    │   │   └── ...\n    │   └── model_pth/\n    │       ├── GIVTEDNet.pth\n    │       └── ...\n    └── WBCImage/\n        ├── TrainDataset/\n        │   ├── images/\n        │   │   ├── image_00.jpg\n        │   │   └── ...\n        │   └── masks/\n        │       ├── image_00.jpg\n        │       └── ...\n        ├── TestDataset/\n        │   ├── images/\n        │   │   ├── image_00.jpg\n        │   │   └── ...\n        │   └── masks/\n        │       ├── image_00.jpg\n        │       └── ...\n        ├── result/\n        │   ├── image_00.jpg\n        │   └── ...\n        └── model_pth/\n            ├── GIVTEDNet.pth\n            └── ...\n```\n\n\n## Training\n\nTo train our model:\n\n```bash\n# Please choose either one: KvasirInstrument, ISIC2018, WBCImage\npython train.py --dataset_name \"KvasirInstrument\" \n```\n\n## Testing\n\nWe conducted the evaluation on the Raspberry Pi 4. However, it can be run on some other hardware, too.\n\n```bash\npython eval.py\n```\n\n\n## Benchmark\n\n### Segmentation Performance\n\n\n![qualitative](./assets/ardiy4-3411870-large.gif)\n\u003cdiv align=\"center\"\u003eThe qualitative results of GIVTED-Net and other models. For quantitative results, please check our paper.\u003c/div\u003e\n\n\n### Computational Performance\n\n| Model  | Parameters  | FLOPs  | FPS  |\n|---|---|---|---|\n| TransUNet  | 105.28M  | 50.70G  |  0.22 |\n| U-Net  | 31.04M  |  83.86G |  0.14 |\n| LeViT-UNet-384  |  52.15M | 50.82G  |  0.26 |\n| LeViT-UNet-192  |  19.89M |  38.58G |  0.36 |\n| LeViT-UNet-128s  |  15.89M | 34.86G  | 0.39  |\n| ResUNet++  | 4.06M  | 24.28G  | 0.31  |\n| SA-UNet  | 0.54M  | 4.86G  |  1.48 |\n|  Mobile-PolypNet | 0.23M  |  2.24G | 0.76  |\n| NanoNet-A  | 0.29M  | 2.44G  | 1.66  |\n|  NanoNet-B | 0.18M  | 1.97G  | 1.71  |\n| NanoNet-C  | 43.34K  | 0.44G  |  3.67 |\n| **GIVTED-Net (Ours)**  | **0.19M**  | **0.56G**  |  **2.47** |\n\n\n## Citation\n\n```bibtex\n@ARTICLE{10552270,\n  author={Dwika Hefni Al-Fahsi, Resha and Naghim Fauzaini Prawirosoenoto, Ahmad and Adi Nugroho, Hanung and Ardiyanto, Igi},\n  journal={IEEE Access}, \n  title={GIVTED-Net: GhostNet-Mobile Involution ViT Encoder-Decoder Network for Lightweight Medical Image Segmentation}, \n  year={2024},\n  volume={12},\n  number={},\n  pages={81281-81292},\n  keywords={Image segmentation;Biomedical imaging;Transformers;Convolutional neural networks;Computational modeling;Decoding;Convolutional neural networks;Deep learning;Deep learning;GhostNet;lightweight model;medical image segmentation;mobile involution ViT},\n  doi={10.1109/ACCESS.2024.3411870}\n}\n```\n\n## Reference\n\n* [KvasirInstrument](https://datasets.simula.no/kvasir-instrument/)\n* [ISIC2018](https://challenge.isic-archive.com/data/#2018)\n* [WBCImage](https://github.com/zxaoyou/segmentation_WBC)\n\n\n\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Freshalfahsi%2Fgivted-net","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Freshalfahsi%2Fgivted-net","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Freshalfahsi%2Fgivted-net/lists"}