Seenka/banners-dict

---
dataset_info:
features:
- name: image
dtype: image
- name: label
dtype:
class_label:
names:
'0': none
'1': videograph
'2': zocalo
- name: cropped_image
dtype: image
- name: embeddings_cropped
sequence: float32
- name: embeddings
sequence: float32
- name: ocr_out
list:
- name: bbox
sequence:
sequence: float64
- name: confidence
dtype: float64
- name: text
dtype: string
splits:
- name: train
num_bytes: 200666273.136
num_examples: 1182
- name: canal_12
num_bytes: 26133565.0
num_examples: 265
download_size: 229251427
dataset_size: 226799838.136
---
# Dataset Card for "banners-dict"

[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)

原始信息汇总

数据集概述

数据集特征

• image: 图像数据
• label: 标签数据，包含以下类别：
  • 0: none
  • 1: videograph
  • 2: zocalo
• cropped_image: 裁剪后的图像数据
• embeddings_cropped: 裁剪后图像的嵌入向量，类型为float32序列
• embeddings: 图像的嵌入向量，类型为float32序列
• ocr_out: OCR输出，包含以下子特征：
  • bbox: 边界框，类型为float64序列的序列
  • confidence: 置信度，类型为float64
  • text: 文本，类型为字符串

数据集分割

• train: 训练集，包含1182个样本，大小为200666273.136字节
• canal_12: 数据集分割，包含265个样本，大小为26133565.0字节

数据集大小

• 下载大小: 229251427字节
• 数据集大小: 226799838.136字节