GangCaoLab/FISH_spots

1---2license: mit3---4 5# FISH_spots Dataset6 7The manually verified in situ hybridization fluorescence images and point coordinate dataset.8This dataset contains images and annotations for the task of single-molecule fluorescence in situ hybridization (FISH) spot detection, supporting 2D, 3D, and simulated noisy data. The structure is designed for deep learning model development, training, and evaluation.9 10## Directory Structure11```12FISH_spots/13├── 2d/14│ ├── csv/15│ ├── image/16│ ├── image_raw/17│ └── meta.csv18├── 3d/19│ ├── csv/20│ ├── image/21│ ├── meta_test.csv22│ └── meta_valid.csv23└── noise/24├── csv/25├── image/26├── meta.csv27├── meta_test.csv28└── meta_valid.csv29 30```31 32## Download33```bash34# Make sure you have git-lfs installed (https://git-lfs.com)35git lfs install36git clone https://huggingface.co/datasets/GangCaoLab/FISH_spots37```38 39## Data Content and Annotation Format40 41- **image directory**  42  Contains 512×512 image patches used for training. The images are ready for use by deep learning models, in standard formats such as PNG or TIFF.43 44- **csv directory**  45  Stores CSV files containing annotated spot positions for each corresponding image patch. Each row records the coordinates and other metadata of detected or simulated spots.46 47- **meta*.csv**  48  `meta.csv`,  provide global metadata.49 50## 2D Dataset51 52- Includes data used in the U-FISH publication, comprising both real experimental data and simulated data.53- Covers 7 different experimental sources, providing diversity (real and simulated conditions).54- Contains over 4,000 images and 160,000 annotated spot positions.55- Suitable for benchmarking model generalization and spot detection in varied scenarios.56 57## 3D Dataset58 59- Mainly generated via simulation, representing spot distributions under 3D imaging conditions.60- Provides 3D image patches and coordinate-based annotations.61- Includes `meta_test.csv` and `meta_valid.csv` for easy split management.62 63## Noise Dataset64 65- Contains simulated noisy images aimed at testing model robustness to noise.66- Structure mirrors the 3D dataset, with patches, annotations, and metadata files.67 68## Citation69 70If you use this dataset in your research, please cite the U-FISH paper and Dataset(DOI: https://doi.org/10.57967/hf/6150).71