CLI Reference

NEMO installs three command-line entry points.

nemo-detect

Run per-channel starlet wavelet detection only.

nemo-detect --cube data/cube.fits --out results/ \
            --scales 6 --k-sigma 5.0 --use-scale 5 --min-area 20

--cube

Input cube file: .fits, .h5, .npy, or .npz.

--out

Output directory for detections_summary.json.

--channels

Comma-separated channel indices. Default: auto-selected active channels.

--scales

Total starlet scales including coarse residual.

--k-sigma

Detection threshold in units of per-scale noise.

--use-scale

1-based detail band used for component extraction.

--min-area

Minimum component area in pixels.

nemo-track

Run the full detection + tracking pipeline.

nemo-track --cube data/cube.fits --out results/ \
           --scales 6 --k-sigma 5.0 --use-scale 5 --min-area 20 \
           --min-match-overlap 5 --min-split-overlap 3 --min-displacement 3.0

Produces tracks.csv, sources.csv, and summary.json in the output directory.

--min-match-overlap

Minimum pixel overlap (advected mask ∩ component) to accept a continuation match.

--min-split-overlap

Minimum pixel overlap to attribute an unmatched component as a split.

--min-displacement

Minimum cumulative centroid travel in pixels to classify a track as kinematic.

nemo-denoise

Denoise a FITS cube with the IST algorithm (requires cosmostat).

nemo-denoise cube.fits --threshold 5.0 --thresh-increm 2.0 --num-iter 20