List tools for datasets creation for speech and computer vision tasks.
- User
List tools for datasets creation for speech and computer vision tasks.
Assistant
Creating datasets for speech and computer vision tasks involves various tools for data collection, annotation, augmentation, and management. Below is a categorized list of popular tools:
1. Speech Dataset Tools
Data Collection & Recording
- Audacity – Open-source audio recording and editing tool.
- OBS Studio – Screen/audio recording for speech data.
- Google Speech Commands Dataset – Pre-recorded datasets for keyword spotting.
- Common Voice (Mozilla) – Crowdsourced multilingual speech dataset.
Annotation & Labeling
- Praat – Phonetic analysis and annotation.
- ELAN – Advanced audio/video annotation.
- Audino – Open-source speech annotation tool.
- Prodigy (by Explosion) – Customizable annotation for ASR/NLP.
Augmentation & Preprocessing
- Librosa (Python) – Audio feature extraction & manipulation.
- TorchAudio (PyTorch) – Audio transformations.
- Audiomentations – Python library for audio augmentation.
Synthetic Speech Generation
- Tacotron 2 (NVIDIA) – Text-to-speech synthesis.
- Coqui TTS – Open-source TTS for dataset generation.
- Google Text-to-Speech (gTTS) – API for synthetic speech.
2. Computer Vision Dataset Tools
Data Collection & Scraping
- OpenCV – Image/video capture and processing.
- Scrapy / BeautifulSoup – Web scraping for images.
- Flickr API / Google Images (with permissions).
- Roboflow Universe – Pre-labeled public datasets.
Annotation & Labeling
- LabelImg – Bounding box annotation (XML/JSON).
- LabelMe (MIT) – Polygonal annotation for segmentation.
- CVAT (Intel) – Advanced video/image annotation.
- Supervisely – Collaborative annotation platform.
- VGG Image Annotator (VIA) – Lightweight web-based tool.
- Prodigy – Custom annotation workflows.
Augmentation & Preprocessing
- Albumentations – Fast image augmentation library.
- imgaug – Python-based augmentation.
- TorchVision (PyTorch) – Built-in transforms.
- Kornia – GPU-accelerated augmentations.
Synthetic Data Generation
- Blender – 3D rendering for synthetic datasets.
- Unity Perception – Generate labeled synthetic CV data.
- NVIDIA Omniverse – Physically accurate synthetic data.
- GANs (StyleGAN, CycleGAN) – Generate realistic images.
3. General Dataset Management
- DVC (Data Version Control) – Track dataset versions.
- Weights & Biases (W&B) / MLflow – Dataset logging.
- Pandas / Dask – Tabular dataset manipulation.
- Hugging Face Datasets – Share/version datasets.
4. Cloud & Auto-Labeling Services
- Amazon SageMaker Ground Truth – Managed labeling.
- Google Vertex AI – AutoML and labeling.
- Scale AI / Labelbox – Enterprise annotation platforms.
- SuperAnnotate – End-to-end CV dataset pipeline.
Choosing the Right Tool
- For small projects: LabelImg (CV) + Audacity (Speech).
- For teams: CVAT (CV) + Prodigy (Speech).
- Synthetic data: Blender (CV) + Coqui TTS (Speech).
Would you like recommendations for a specific use case (e.g., autonomous driving, medical imaging, or multilingual ASR)?
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.