Update README.md

README.md

@@ -56,12 +56,9 @@ dataset_summary: '
   '
 ---
 
-# Dataset Card for hand_keypoints
-
-<!-- Provide a quick summary of the dataset. -->
-
-
+# Dataset Card for Image Hand Keypoint Detection
 
+![image/png](hands-dataset.gif)
 
 
 This is a [FiftyOne](https://github.com/voxel51/fiftyone) dataset with 846 samples.
@@ -91,40 +88,26 @@ session = fo.launch_app(dataset)
 
 ## Dataset Details
 
-### Dataset Description
-
-
-![image/png](hands-dataset.gif)
+As part of their research, the authors created a dataset by manually annotating two publicly available image sets: the MPII Human Pose dataset and images from the New Zealand Sign Language (NZSL) Exercises. To date, they collected annotations for 1300 hands on the MPII set and 1500 on NZSL. This combined dataset was split into a training set (2000 hands) and a testing set (800 hands)
 
 
+### Dataset Description
 
+The dataset created in this research is a collection of manually annotated RGB images of hands sourced from the MPII Human Pose dataset and the New Zealand Sign Language (NZSL) Exercises. It contains 2D locations for 21 keypoints on 2800 hands, split into a training set of 2000 hands and a testing set of 800 hands. This dataset was used to train and evaluate their hand keypoint detection methods for single images.
 - **Curated by:** Tomas Simon, Hanbyul Joo, Iain Matthews, Yaser Sheikh
 - **Funded by:** Carnegie Mellon University
 - **Shared by:** [Harpreet Sahota](https://huggingface.co/harpreetsahota), Hacker-in-Residence at Voxel51
 - **License:** [More Information Needed]
 
 ### Dataset Sources
-
-As part of their research, the authors created a dataset by manually annotating two publicly available image sets: the MPII Human Pose dataset and images from the New Zealand Sign Language (NZSL) Exercises. To date, they collected annotations for 1300 hands on the MPII set and 1500 on NZSL. This combined dataset was split into a training set (2000 hands) and a testing set (800 hands)
-
 - **Paper:** https://arxiv.org/abs/1704.07809
 - **Demo:** http://domedb.perception.cs.cmu.edu/handdb.html
 
 ## Uses
 
-<!-- Address questions around how the dataset is intended to be used. -->
-
 ### Direct Use
 
-<!-- This section describes suitable use cases for the dataset. -->
-
-[More Information Needed]
-
-### Out-of-Scope Use
-
-<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->
-
-[More Information Needed]
+This manually annotated dataset was directly used to train and evaluate their hand keypoint detection methods. The dataset serves as a benchmark to assess the accuracy of their single image 2D hand keypoint detector. It enabled them to train an initial detector and evaluate the improvements gained through their proposed multiview bootstrapping technique. The dataset contains images extracted from YouTube videos depicting everyday human activities (MPII) and images showing a variety of hand poses from people using New Zealand Sign Language (NZSL). These diverse sets of images allowed the researchers to evaluate the generalization capabilities of their detector.
 
 ## Dataset Structure
 
@@ -146,64 +129,27 @@ Sample fields:
     left_hand:        fiftyone.core.fields.EmbeddedDocumentField(fiftyone.core.labels.Keypoints)
 ```
 
-
-## Dataset Creation
-
-### Curation Rationale
-
-<!-- Motivation for the creation of this dataset. -->
-
-[More Information Needed]
-
 ### Source Data
 
-<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->
-
 #### Data Collection and Processing
 
-<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->
+The dataset is composed of manually annotated RGB images of hands sourced from two existing datasets: MPII and NZSL.
 
-[More Information Needed]
+• **Annotations:** Each annotated image includes 2D locations for 21 keypoints on the hand (see Fig. 4a for an example). These keypoints represent different landmarks on the hand, such as finger tips and joints.
 
-#### Who are the source data producers?
+• **Splits:** The combined dataset of 2800 annotated hands was divided into a training set of 2000 hands and a testing set of 800 hands. The criteria for this split are not explicitly detailed in the provided excerpts.
 
-<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->
+Source Datasets:
 
-[More Information Needed]
+◦ **MPII Human Pose dataset:** Contains images from YouTube videos depicting a wide range of everyday human activities. These images vary in quality, resolution, and hand appearance, and include various types of occlusions and hand-object/hand-hand interactions.
+◦ **New Zealand Sign Language (NZSL) Exercises:** Features images of people making visible hand gestures for communication. This subset provides a variety of hand poses commonly found in conversational contexts
 
-### Annotations [optional]
-
-<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->
+### Annotations
 
 #### Annotation process
 
-<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
-
-[More Information Needed]
-
-#### Who are the annotators?
-
-<!-- This section describes the people or systems who created the annotations. -->
-
-[More Information Needed]
-
-#### Personal and Sensitive Information
-
-<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->
-
-[More Information Needed]
-
-## Bias, Risks, and Limitations
-
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
-
-[More Information Needed]
-
-### Recommendations
-
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
-
-Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
+This manually annotated dataset was directly used to train and evaluate their hand keypoint detection methods. The dataset serves as a benchmark to assess the accuracy of their single image 2D hand keypoint detector. It enabled them to train an initial detector and evaluate the improvements gained through their proposed multiview bootstrapping technique. The dataset contains images extracted from YouTube videos depicting everyday human activities (MPII) and images showing a variety of hand poses from people using New Zealand Sign Language (NZSL). These diverse sets of images allowed the researchers to evaluate the generalization capabilities of their detector
+The process of manually annotating hand keypoints in single images was challenging due to frequent occlusions caused by hand articulation, viewpoint, and grasped objects (as illustrated in Fig. 2). In many cases, annotators had to estimate the locations of occluded keypoints, potentially reducing the accuracy of these annotations
 
 ## Citation