25 datasets found

The dataset used in the paper is a simulated dough manipulation environment, where the goal is to create a donut, a baguette, and two pancakes using a set of candidate tools.

Semantic parsers map natural language utterances into meaning representations (e.g. programs). Such models are typically bottle-necked by the paucity of training data due to the...

The UT-Zappos50K dataset is a fine-grained shoe catalog, characterized by its smaller scale and relatively stable and simple content.

Finetuned language models are zero-shot learners

Zero-shot one image to 3D object.

Text-conditioned human motion generation has experienced significant advancements with diffusion models trained on extensive motion capture data and corresponding textual...

The dataset used for evaluation of the MS-CLIP model, which consists of 25 public datasets for zero-shot learning and linear probing.

Semantic segmentation performs pixel-level classifica- tion to localize objects from different classes in the input image. Open-vocabulary semantic segmentation aims to...

Google Open Images dataset, which contains 19,958 categories and is used for zero-shot learning.

Zero-shot video question answering via frozen bidirectional language models.

AWA2 is an animal dataset containing 37,322 images from 50 classes, with 85 attributes provided by experts to describe the semantic feature of each class.

The dataset used in the paper is HMDB51 and UCF101.

The dataset used in the paper is Kinetics-400 and Something-Something-V2.

Given an untrimmed video and a language query, video grounding aims to localize the time interval by understanding the text and video simultaneously.

A custom dataset to simulate a user's image directory for testing the ZeroSearch algorithm.

The dataset used in the paper is a multi-view clustering dataset, which contains 6 views of 30000 samples each. The dataset is used to evaluate the performance of the proposed...

VGDiffZero is a zero-shot visual grounding framework that leverages pre-trained text-to-image diffusion models' vision-language alignment abilities.

The Kinetics-600 dataset consists of 392k training videos and 30k validation videos in 600 human action categories.

Motion has shown to be useful for video understanding, where motion is typically represented by optical flow. However, computing flow from video frames is very time-consuming....

Motion has shown to be useful for video understanding, where motion is typically represented by optical flow. However, computing flow from video frames is very time-consuming....