joseluhf11/oct-object-detection-v2-average

---
dataset_info:
features:
- name: image
dtype: image
- name: objects
struct:
- name: bbox
sequence:
sequence: float64
- name: categories
sequence: string
splits:
- name: train
num_bytes: 153967507.25
num_examples: 1246
download_size: 71639902
dataset_size: 153967507.25
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---
# Dataset Card for "oct-object-detection-v2-average"
Dataset is composed of images with multiples object detection box in coco format (x,y,w,h). Images are OCT (type of eye scaner) with boxes indicating some features associated to AMD disease.
Changes from from v1 are images are grouped into a single row for the same class detection object, and also join with average method overlapping boxes. Average means, average of Xs and Ys.
[Source datataset](https://doi.org/10.1101/2023.03.29.534704)

原始信息汇总

数据集概述

数据集信息

• 特征:

  • image: 图像数据类型
  • objects: 结构化数据，包含以下字段:
    • bbox: 序列类型，序列元素为浮点数
    • categories: 序列类型，序列元素为字符串

• 数据分割:

  • train: 训练集，包含1246个样本，总大小为153967507.25字节

• 数据大小:

  • 下载大小: 71639902字节
  • 数据集大小: 153967507.25字节

配置信息

• 默认配置:
  • config_name: default
  • data_files:
    • split: train
    • path: data/train-*

数据集描述

• 数据集包含图像，图像中包含多个对象检测框（COCO格式：x, y, w, h）。这些图像是OCT（一种眼科扫描图像），检测框指示与AMD疾病相关的特征。
• 与v1版本相比，v2版本中图像被分组为同一类别检测对象的单行，并且使用平均方法合并重叠的检测框。平均方法指的是X和Y的平均值。