neurocog-rime-core

neurocog-rime-core is the headless domain layer for RIME. It provides session models, protocol schemas, annotation storage, rule evaluation, signal loading, ELAN import, export utilities, IRR/coverage/evaluation metrics, and CMF-based model inference helpers.

Install

pip install neurocog-rime-core

For local development:

pip install -e packages/rime-core

Optional extras:

pip install -e "packages/rime-core[onnx,video]"

Quick Start

from pathlib import Path

from rime_core.annotation import AnnotationStore
from rime_core.schema import ProtocolSchema
from rime_core.sessions import VideoConfig, create_session
from rime_core.workspace import WorkingContext

session = create_session(
    session_dir=Path("example-session"),
    name="Example Session",
    videos=[VideoConfig(path="video.mp4", role="primary")],
)

context = WorkingContext.open(session.session_dir)
schema = ProtocolSchema.default()
store = AnnotationStore()

print(session.name)
print(context.session.session_dir)
print(schema.get_lane_names()[:3])
print(len(store.all()))

Main Modules

• rime_core.annotation: annotations, review, and rule helpers
• rime_core.analysis: coverage, IRR, and model-evaluation utilities
• rime_core.io: import/export and signal-loading helpers
• rime_core.modeling: CMF package loading and inference
• rime_core.sessions: session dataclasses and persistence
• rime_core.workspace: live session orchestration