Update README.md

README.md

@@ -3,72 +3,22 @@ annotations_creators: []
 language: en
 license: mit
 size_categories:
-- 10K<n<100K
+  - "10K<n<100K"
 task_categories:
-- image-classification
+  - image-classification
 task_ids: []
 pretty_name: asl-mnist
 tags:
-- fiftyone
-- image
-- image-classification
+  - fiftyone
+  - image
+  - image-classification
 repo_type: dataset
-dataset_summary: '
-
-
-
-
-  This is a [FiftyOne](https://github.com/voxel51/fiftyone) dataset with 34627 samples.
-
-
-  ## Installation
-
-
-  If you haven''t already, install FiftyOne:
-
-
-  ```bash
-
-  pip install -U fiftyone
-
-  ```
-
-
-  ## Usage
-
-
-  ```python
-
-  import fiftyone as fo
-
-  from fiftyone.utils.huggingface import load_from_hub
-
-
-  # Load the dataset
-
-  # Note: other available arguments include ''max_samples'', etc
-
-  dataset = load_from_hub("andandandand/American-Sign-Language-MNIST")
-
-
-  # Launch the App
-
-  session = fo.launch_app(dataset)
-
-  ```
-
-  '
+dataset_summary: "This is a FiftyOne dataset with 34,627 samples of American Sign Language (ASL) alphabet images converted from the original Kaggle Sign Language MNIST dataset. The dataset contains 28x28 pixel grayscale images of hand gestures representing 24 letters of the ASL alphabet (excluding J and Z which require motion)."
 ---
 
-# Dataset Card for asl-mnist
-
-<!-- Provide a quick summary of the dataset. -->
-
+# Dataset Card for ASL-MNIST
 
-
-
-
-This is a [FiftyOne](https://github.com/voxel51/fiftyone) dataset with 34627 samples.
+This is a [FiftyOne](https://github.com/voxel51/fiftyone) dataset with 34,627 samples of American Sign Language (ASL) alphabet images, converted from the original Kaggle Sign Language MNIST dataset into a format optimized for computer vision workflows.
 
 ## Installation
 
@@ -92,135 +42,167 @@ dataset = load_from_hub("andandandand/American-Sign-Language-MNIST")
 session = fo.launch_app(dataset)
 ```
 
-
 ## Dataset Details
 
 ### Dataset Description
 
-<!-- Provide a longer summary of what this dataset is. -->
+This dataset contains 34,627 grayscale images of American Sign Language (ASL) alphabet hand gestures. Each image is 28x28 pixels and represents one of 24 letters from the ASL alphabet. The dataset excludes the letters J and Z as these require motion to be properly signed and cannot be represented in static images.
 
+The original data was sourced from Kaggle's Sign Language MNIST dataset and has been converted into individual image files with proper FiftyOne annotations for easy visualization, querying, and analysis.
 
-
-- **Curated by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
+- **Curated by:** Antonio Rueda-Toicen (Voxel51)
+- **Funded by:** Voxel51
+- **Shared by:** Voxel51
 - **Language(s) (NLP):** en
-- **License:** mit
-
-### Dataset Sources [optional]
+- **License:** MIT
 
-<!-- Provide the basic links for the dataset. -->
+### Dataset Sources
 
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
+- **Original Repository:** [Kaggle Sign Language MNIST](https://www.kaggle.com/datasets/datamunge/sign-language-mnist)
+- **Processing Code:** [Google Colab Notebook](https://github.com/andandandand/practical-computer-vision/blob/main/ASL_from_Kaggle_to_FiftyOne.ipynb) - Converted from CSV format to individual JPG images using FiftyOne
+- **FiftyOne Documentation:** [https://github.com/voxel51/fiftyone](https://github.com/voxel51/fiftyone)
 
 ## 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]
+This dataset is ideal for:
+- **Computer vision research**: Training and evaluating image classification models
+- **Sign language recognition**: Developing ASL alphabet recognition systems
+- **Educational purposes**: Teaching machine learning concepts with a meaningful, accessible dataset
+- **Prototyping**: Quick experimentation with image classification algorithms
+- **FiftyOne demonstrations**: Showcasing dataset visualization and management capabilities
 
 ### Out-of-Scope Use
 
-<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->
-
-[More Information Needed]
+This dataset should not be used for:
+- **Complete ASL communication systems**: Missing J and Z letters, and doesn't include words, phrases, or grammar
+- **Real-time sign language interpretation**: Static images don't capture the dynamic nature of full ASL communication
+- **Medical or accessibility applications**: Without proper validation in clinical settings
+- **Production ASL recognition systems**: Without additional data augmentation and validation
 
 ## Dataset Structure
 
-<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
+The dataset contains 34,627 samples split into training and test sets:
+- **Training samples**: ~27,455 images
+- **Test samples**: ~7,172 images
+- **Image format**: 28x28 pixel grayscale JPG images
+- **Labels**: 24 ASL alphabet letters (A-I, K-Y, excluding J and Z)
 
-[More Information Needed]
+### Label Mapping
+
+The dataset uses the following label mapping:
+```
+0: 'A', 1: 'B', 2: 'C', 3: 'D', 4: 'E', 5: 'F', 6: 'G', 7: 'H', 8: 'I',
+10: 'K', 11: 'L', 12: 'M', 13: 'N', 14: 'O', 15: 'P', 16: 'Q', 17: 'R',
+18: 'S', 19: 'T', 20: 'U', 21: 'V', 22: 'W', 23: 'X', 24: 'Y'
+```
+
+Note: Labels 9 (J) and 25 (Z) are excluded as these letters require motion in ASL.
 
 ## Dataset Creation
 
 ### Curation Rationale
 
-<!-- Motivation for the creation of this dataset. -->
-
-[More Information Needed]
+This dataset was created to provide an easily accessible, well-structured version of the ASL alphabet dataset for computer vision research and education. The conversion to FiftyOne format enables:
+- Rich visualization and exploration capabilities
+- Advanced querying and filtering
+- Seamless integration with machine learning workflows
+- Better understanding of dataset characteristics and potential biases
 
 ### 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. -->
-
-[More Information Needed]
+The original data came from the Kaggle Sign Language MNIST dataset. The processing pipeline included:
 
-#### Who are the source data producers?
+1. **Download**: Retrieved CSV files from Kaggle using the Kaggle API
+2. **Image conversion**: Converted pixel arrays from CSV rows into 28x28 grayscale images
+3. **File organization**: Saved images as individual JPG files with systematic naming
+4. **Annotation**: Created FiftyOne classification labels for each image
+5. **Metadata computation**: Added image metadata for enhanced searchability
 
-<!-- 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. -->
+Processing was performed using Python with pandas, PIL, NumPy, and FiftyOne libraries.
 
-[More Information Needed]
+#### Who are the source data producers?
 
-### Annotations [optional]
+The original dataset was created by the Kaggle user 'datamunge' and represents hand gesture images collected for ASL alphabet recognition research. The specific methodology for the original data collection is not detailed in the source.
 
-<!-- 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]
+Annotations were automatically generated during the conversion process:
+- Each image received a classification label corresponding to its ASL letter
+- Labels were mapped from numeric indices to letter strings
+- Split information (train/test) was preserved as tags
+- No manual annotation was required as labels were inherited from the original dataset
 
 #### Who are the annotators?
 
-<!-- This section describes the people or systems who created the annotations. -->
-
-[More Information Needed]
-
-#### Personal and Sensitive Information
+The original labels were provided in the source Kaggle dataset. The FiftyOne conversion and annotation formatting was performed by Antonio Rueda-Toicen at Voxel51.
 
-<!-- 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. -->
+### Personal and Sensitive Information
 
-[More Information Needed]
+The dataset contains images of human hands performing sign language gestures. While hands are visible, no faces or other personally identifiable information is included in the images. The 28x28 pixel resolution is too low to identify individuals from hand features alone.
 
 ## Bias, Risks, and Limitations
 
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+### Known Limitations
 
-[More Information Needed]
+- **Incomplete alphabet**: Missing J and Z letters limits complete ASL alphabet coverage
+- **Static images only**: Cannot represent the dynamic/motion aspects of ASL
+- **Limited diversity**: May not represent the full diversity of hand sizes, skin tones, or signing styles
+- **Low resolution**: 28x28 pixels may not capture fine details important for real-world applications
+- **Single signer perspective**: May not generalize well to different signing styles or hand appearances
 
-### Recommendations
+### Potential Biases
 
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+- **Demographic representation**: Unknown diversity in the original data collection
+- **Lighting conditions**: Images may reflect specific lighting setups that don't generalize
+- **Hand positioning**: May favor certain hand orientations or positions
+- **Background consistency**: Controlled backgrounds may not reflect real-world conditions
 
-Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
+### Recommendations
 
-## Citation [optional]
+Users should be aware of these limitations and consider:
+- Supplementing with additional diverse ASL datasets for production use
+- Validating model performance across different demographic groups
+- Testing with real-world conditions and lighting variations
+- Incorporating motion-based data for complete ASL recognition systems
+- Consulting with ASL community members for culturally appropriate applications
 
-<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->
+## Citation
 
 **BibTeX:**
+```bibtex
+@dataset{rueda_toicen_2024_asl_mnist,
+  title={American Sign Language MNIST Dataset (FiftyOne Format)},
+  author={Rueda-Toicen, Antonio},
+  year={2024},
+  publisher={Hugging Face},
+  url={https://huggingface.co/datasets/andandandand/American-Sign-Language-MNIST}
+}
+```
 
-[More Information Needed]
-
-**APA:**
-
-[More Information Needed]
-
-## Glossary [optional]
-
-<!-- If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->
+## Glossary
 
-[More Information Needed]
+- **ASL**: American Sign Language, a complete, natural language that uses hand gestures, facial expressions, and body language
+- **FiftyOne**: An open-source tool for building high-quality datasets and computer vision models
+- **MNIST**: A reference to the famous Modified National Institute of Standards and Technology database, indicating 28x28 pixel grayscale images
+- **Classification**: The machine learning task of assigning a category or label to an input
 
-## More Information [optional]
+## More Information
 
-[More Information Needed]
+For more information about FiftyOne and dataset management best practices, visit:
+- [FiftyOne Documentation](https://docs.voxel51.com/)
+- [FiftyOne GitHub Repository](https://github.com/voxel51/fiftyone)
+- [Computer Vision Datasets Guide](https://docs.voxel51.com/user_guide/dataset_creation/index.html)
 
-## Dataset Card Authors [optional]
+## Dataset Card Authors
 
-[More Information Needed]
+Antonio Rueda-Toicen (Voxel51)
 
 ## Dataset Card Contact
 
-[More Information Needed]
+Antonio Rueda-Toicen - [email protected]