Papers with Code · CC BY-SA 4.0

데이터셋 탐색

데이터셋의 설명·홈페이지를 찾고, 관련 태스크의 평가 기록을 살펴보세요.

관련 논문 18편

HOST

HOST 설명 보기

The heavily occluded scene text (HOST) dataset is a dataset that contains images of text with occlusions. It is used to improve the recognition performance of occluded text in machine vision applications 1. The dataset is composed of 4832 images that are manually occluded in weak or heavy degrees.

관련 논문 13편

Ghostbuster

Ghostbuster 설명 보기

The Ghostbusters dataset leverages the GPT-3.5-turbo model for generating texts in the domains of creative writing, news, and student essays, providing 2,000 texts in the first two domains and 1,994 in the latter.

관련 논문 7편

GHOSTS

GHOSTS 설명 보기

GHOSTS is the first natural-language dataset made and curated by working researchers in mathematics that (1) aims to cover graduate-level mathematics and (2) provides a holistic overview of the mathematical capabilities of language models. It a collection of multiple datasets of prompts, totalling 728 prompts, for which ChatGPT’s output was manually rated by experts. Source: Mathematical Capabilities of ChatGPT

관련 논문 0편

ZakynthosTurtles Dataset

ZakynthosTurtles Dataset 설명 보기

Description: The ZakynthosTurtles dataset has been designed to support the development of numerical methods for the recognition and re-identification of individual sea turtles based on their unique scale patterns. Each turtle’s distinct scale arrangement provides a natural identifier, allowing researchers to track and study individuals over time. This dataset contains photographs captured in the wild, emphasizing the importance of non-intrusive wildlife monitoring techniques. The dataset includes 40 individual loggerhead sea turtles, with each turtle represented by four high-resolution photographs—both left and right profiles taken in two different years. This structure offers unique opportunities for investigating the stability and similarity of scale patterns over time and between opposite profiles. Download Dataset Dataset Structure: Species: Loggerhead sea turtles (Caretta caretta) Total Turtles: 40 individuals Photos per Turtle: 4 (Left and right profile taken in two different years) Total Photos: 160 high-quality images taken underwater This structured dataset enables scientists to study important questions in computer vision and machine learning, such as cross-year recognition and left-right profile matching, offering insights into how re-identification algorithms handle natural variations in appearances over time. Associated Datasets: The ZakynthosTurtles dataset is part of a larger family of sea turtle datasets, each contributing unique insights and use cases. These include: AmvrakikosTurtles: A public dataset comprising 50 loggerhead sea turtles, captured from a boat during sea surveys. This dataset primarily focuses on surface-level photography. ReunionTurtles: This dataset features 50 green and 34 hawksbill sea turtles, photographed underwater, providing additional variety in species and underwater imaging conditions. Together, these datasets provide a comprehensive foundation for designing and testing machine learning models for sea turtle identification and re-identification. Related Resources: wildlife-datasets: A Python library summarizing various public datasets for wildlife re-identification, including additional turtle datasets. wildlife-tools: A Python library offering tools for training and testing re-identification models using wildlife data. Research Paper 1: This paper covers our work with another turtle dataset, featuring 438 individual sea turtles, further expanding the scope of research. Research Paper 2: This paper details the development and application of the Python packages mentioned above, aimed at enhancing re-identification workflows. Conclusion: The ZakynthosTurtles dataset provides a valuable resource for advancing research in wildlife re-identification, particularly for sea turtles. Its structured format and multi-year imagery make it an excellent tool for researchers and developers working in the fields of machine learning, computer vision, and wildlife monitoring. By contributing to the refinement of individual recognition algorithms, this dataset plays a key role in non-invasive wildlife management and conservation efforts, particularly for endangered species like the loggerhead sea turtles. This dataset is sourced from Kaggle.