Data & MLOps
Ultrasound Video → Dataset Pipeline
Deepecho
Turns large volumes of raw ultrasound studies into clean, annotated, ML-ready datasets.
- →Video decoding, frame extraction and segmentation into fixed-length clips
- →Model-assisted pre-annotation in CVAT, exported to COCO
- →Deduplication, quality control, and separate noisy / cleaned dataset versions
- →Metadata extraction and storage on Google Cloud Storage, processed with Dataflow
Data EngineeringCVATCOCODataflowGCSVideo Processing