README.md

---
language:
- en
size_categories:
- 10K<n<100K
task_categories:
- object-detection
pretty_name: BF Microscopy SCC Filtered
dataset_info:
 features:
 - name: image
 dtype: image
 - name: label
 dtype:
 class_label:
 names:
 '0': test
 '1': train
 '2': val
 - name: width
 dtype: int64
 - name: height
 dtype: int64
 - name: objects
 struct:
 - name: bbox
 sequence:
 sequence: float64
 - name: categories
 sequence: int64
 - name: well_edge
 dtype: bool
 splits:
 - name: train
 num_bytes: 4938750495.413194
 num_examples: 20558
 - name: validation
 num_bytes: 608027081.3190594
 num_examples: 2527
 - name: test
 num_bytes: 288591367.8835443
 num_examples: 1200
 download_size: 5795214422
 dataset_size: 5835368944.615797
configs:
- config_name: default
 data_files:
 - split: train
 path: data/train-*
 - split: validation
 path: data/validation-*
 - split: test
 path: data/test-*
tags:
- biology
---
# Brightfield Microscopy SCC Dataset
Filtered to only contain images with no well edge visible

## Dataset Description

The Brightfield Microscopy SCC Dataset contains brightfield microscopy images that have been sliced from 4K resolution images. The cells in the images are labeled with bounding boxes. This dataset is used for training unconditional diffusion models and cell detection models.

## Dataset Structure

The dataset is organized into the following splits:

- `train`: 20,558 examples
- `validation`: 2,527 examples 
- `test`: 1,199 examples

### Data Fields

- `image`: The brightfield microscopy image (image)
- `label`: The split the image belongs to (class label)
 - `0`: test
 - `1`: train 
 - `2`: val
- `width`: The width of the image (int64)
- `height`: The height of the image (int64) 
- `objects`: A struct containing:
 - `bbox`: A sequence of sequences representing the bounding box coordinates (float)
 - `categories`: A sequence of integers representing the object categories (int)
- `well_edge`: Whether this image contains elements of a microtiter plate well edge (bool)