HMDB51 Dataset

Estimated reading: 3 minutes 157 views

Visualization of the HMDB51 Train Dataset on the Deep Lake UI

HMDB51 dataset

What is HMDB51 Dataset?

The HMDB51 (Human Motion Database 51) dataset is created to enhance the research in computer vision research of recognition and search in video. A lot of effort has been put into collection and annotation of large scalable static images with large image categories, but similar effort has not been done in video division. This dataset has been collected to push the efforts in collection and annotation of video datasets. This dataset is a collection of various sources such as movies, public databases (Prelinger archive, YouTube and Google videos). This dataset contains minimum of 101 clips for each 51 action categories and in total, the dataset contains 6849 clips.

Download HMDB51 Dataset in Python

Instead of downloading the HMDB51 dataset in Python, you can effortlessly load it in Python via our Deep Lake open-source with just one line of code.

Load HMDB51 Dataset Training Subset in Python

				
					import deeplake
ds = deeplake.load("hub://activeloop/hmdb51-train")
				
			

Load HMDB51 Dataset Testing Subset in Python

				
					import deeplake
ds = deeplake.load("hub://activeloop/hmdb51-test")
				
			

Load HMDB51 Dataset Extras Subset in Python

				
					import deeplake
ds = deeplake.load("hub://activeloop/hmdb-extras")
				
			

HMDB51 Dataset Structure

HMDB51 Data Fields
  • videos: tensor containing videos
  • labels: tensor containing labels for their respective videos
  • video_quality: tensor containing video quality label
  • number_of_people: tensor containing number of people in a video
  • camera_viewpoint: tensor containing camera viewpoint label for a video
  • camera_motion: tensor containing camera motion label for a video
  • visible_body_parts: tensor containing visible body parts label for a video

How to use HMDB51 Dataset with PyTorch and TensorFlow in Python

Train a model on HMDB51 dataset with PyTorch in Python

Let’s use Deep Lake built-in PyTorch one-line dataloader to connect the data to the compute:

				
					dataloader = ds.pytorch(num_workers=0, batch_size=4, shuffle=False)
				
			
Train a model on HMDB51 dataset with TensorFlow in Python
				
					dataloader = ds.tensorflow()
				
			

Additional Information about HMDB51 Dataset

HMDB51 Dataset Description

HMDB51 Dataset Curators

H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre

HMDB51 Dataset Licensing Information
HMDB51 Dataset Citation Information
				
					@InProceedings{Kuehne11,
   author= "Kuehne, H. and Jhuang, H. and Garrote, E. and Poggio, T. and Serre, T.",
   title = "{HMDB}: a large video database for human motion recognition",
   booktitle = "Proceedings of the International Conference on Computer Vision (ICCV)",
   year = "2011",
}
				
			
CONTENTS