Results for “nvidia-tao-daft”

Validates NVIDIA TAO DAFT datasets for structure, schema, and cross-reference errors using the `tao-daft validate` CLI tool.

Accelerate pandas workflows with GPU DataFrames using cuDF and dask-cuDF for ETL, joins, groupby, and large-scale data processing.

Build video analytics pipelines using NVIDIA DeepStream SDK 9.0 with Python pyservicemaker API, including GStreamer-based video processing, TensorRT inference integration, object detection/tracking, and Kafka/message broker integration.

Performs deep root cause analysis on NVIDIA TAO Visual ChangeNet classification experiments, using image-evidence-driven investigation to diagnose model failures and produce actionable reports.

Embeds target and source image parquets, then mines nearest-neighbour source images for augmentation in VCN AOI workflows.

GPU-accelerated data curation for LLM training, supporting text, image, video, and audio with fuzzy deduplication, quality filtering, semantic deduplication, PII redaction, and NSFW detection.