s6.app.data package¶

Submodules¶

s6.app.data.collect module¶

Collect a random-sweep dataset of synchronized camera frames.

Moves servos randomly within a circular region and captures live camera snapshots via HTTP streaming. Stores each frame in a s6.utils.StructuredDataset with camera ordering resolved by hardware IDs.

Examples

Collect 100 frames with a 5.0 radius and save to a folder:

>>> python -m s6.app.data.collect -n 100 -r 5.0 ...     --waypoint center -o data/random_dataset
s6.app.data.collect.main()¶

Run servo random sweep and capture synchronized snapshots.

s6.app.data.collect.parse_args()¶

Parse CLI arguments for random sweep capture.

s6.app.data.convert module¶

Convert a structured annotation dataset into YOLO format.

Class labels correspond to vertex names (one bounding box per projected vertex). Reads annotation.jsonl and images from the input directory, generates train/val splits, copies images to images/{train,val}, writes YOLO label files, and creates data.yaml for Ultralytics YOLO training.

Examples

>>> python -m s6.app.data.convert -i ./dataset1 -o ./dataset1_yolo --val-ratio 0.2
s6.app.data.convert.collect_classes(ann_file)¶

Collect ordered class names from vertex labels appearing in frames.

s6.app.data.convert.load_annotations(ann_file)¶

Yield parsed InstrumentTrackingFrame entries from a JSONL file.

s6.app.data.convert.main()¶

Run the conversion and write YOLO dataset structure to output.

s6.app.data.convert.parse_args()¶

Parse CLI options for dataset conversion to YOLO format.

s6.app.data.filter module¶

Interactively filter entries in a StructuredDataset.

Displays the image at a specified datakey and overlays a tip point. Use a/d to navigate, x to delete the current entry, and q to quit. A zoomed crop around the keypoint is shown for precision.

s6.app.data.filter.get_by_datakey(data, datakey)¶

Retrieve a nested value using a slash-separated datakey path.

Parameters:

  • data (Mapping | object) – Nested structure (dict-like or object with attributes).

  • datakey (str) – Slash-separated path, e.g., "B/image" or "B/tip_point".

Returns:

The resolved value.

Return type:

Any

s6.app.data.filter.main()¶

Open an interactive viewer to browse and prune dataset entries.

s6.app.data.instrument_render_v3 module¶

s6.app.data.list module¶

Query, count, or download dataset records via the Gateway API.

Builds a Mongo-style aggregation pipeline from simple filter expressions and either prints matching documents, outputs a count, or downloads data into a s6.utils.StructuredDataset.

Examples

  • Count all valid records with score > 0.8:

    >>> python -m s6.app.data.list valid=true score>0.8 --count

  • Download into a local dataset directory:

    >>> python -m s6.app.data.list type="instrument" -o ./out_dataset
s6.app.data.list.main()¶

Parse filters and run query/count/download against the Gateway.

s6.app.data.render module¶

Run Blender to render instrument datasets using a bundled project.

This wrapper launches Blender in background mode with the provided instrument_render_v3.blend and executes the bpy script instrument_render_v3.py. Extra arguments after -- are forwarded to the bpy script.

Examples

Render two loops, forwarding target selection to the bpy script:

>>> python -m s6.app.data.render --loop 2 -- --targets TipTool Grasper
s6.app.data.render.main()¶

Parse CLI flags and invoke Blender with the project and bpy script.

s6.app.data.upload module¶

Upload a dataset via Gateway to a database collection.

Reads a JSONL file (data.jsonl) and associated image folders under a dataset directory, loads each sample into memory, and uploads them to a Gateway server database. Adds metadata fields (upload_datetime, dataset_name, plus any custom KEY=VALUE pairs) to each record.

Examples

>>> python -m s6.app.data.upload ./data/my_dataset -n MyDataset ...   -m split="train" -m tag="v1" -c ./configs/db.config.json
s6.app.data.upload.main()¶

Load samples from disk and upload them via Gateway.

s6.app.data.upload.parse_args()¶

Parse CLI arguments for dataset upload.

s6.app.data.visualize module¶

Preview dataset frames with annotations.

Supports two formats: - InstrumentTrackingFrame with projections and optional boxes. - YOLO layout under images/ and labels/ with .txt labels.

Examples

>>> python -m s6.app.data.visualize ./dataset_dir
s6.app.data.visualize.draw_instrument_frame(img, frame: InstrumentTrackingFrame)¶

Draw instrument projections or boxes onto an image inplace.

s6.app.data.visualize.draw_yolo_frame(img, frame: YoloFrame)¶

Draw YOLO-format boxes (xc,yc,w,h normalized) on an image inplace.

s6.app.data.visualize.load_instrument_frames(ann_path: Path)¶

Load InstrumentTrackingFrame entries from a JSONL file.

s6.app.data.visualize.load_yolo_frames(dataset_dir: Path)¶

Load YOLO frames from an images/train tree with labels.

s6.app.data.visualize.main()¶

Open an interactive viewer to step through annotated frames.

Module contents¶