What problem does it solve? Managing .dxnn model files for DX-Stream GStreamer pipelines requires knowing which models are available, which postprocess library each model needs, and how to verify a model works before wiring it into a pipeline. This Skill centralizes that registry knowledge and validation workflow. ## Core Features & Use Cases - Model Registry Queries: Look up the 14 supported .dxnn models by task type (detection, face detection, pose estimation, segmentation, classification) using the model_list.json schema. - Model Download: Fetch specific models or the full set via setup.sh into the dx_stream samples/models directory. - Pipeline Validation: Verify a .dxnn file, its matching libpostprocess shared library, and exported PostProcess symbol, then test with a minimal gst-launch-1.0 pipeline. - Use Case: You want to run YOLOv8N object detection on a DEEPX NPU. Use this Skill to confirm the model is in the registry, download it, locate libpostprocess_yolov8n.so, and validate the full dxpreprocess → dxinfer → dxpostprocess chain before building your application. ## Quick Start Ask the agent to check whether the YoloV8N model is available in the registry and validate it for use in a dx_stream pipeline.
