364d4de96f
docs(mermaid): fixing mermaid diagram
2026-06-16 18:02:59 +02:00
41a942658b
fix(pyav check): fixing PyAV option validation for integer codec options by normalizing
...
numeric values before calling `is_integer()`
Co-authored-by: Wensi (Vince) Ai <59036629+wensi-ai@users.noreply.github.com >
2026-06-16 18:02:59 +02:00
030d9a279a
chore(format): formatting code
2026-06-16 18:02:59 +02:00
30479cf277
test(dataset tools): adding missing tests for new dataset edition tools features
2026-06-16 18:02:58 +02:00
cb6b2d77bd
test(fix): fixing depth tests
2026-06-16 18:02:58 +02:00
76f79f3955
docs(depth): improving depth maps docs
2026-06-16 18:02:58 +02:00
9e994baa04
chore(format): formatting code
2026-06-16 18:02:56 +02:00
6fd911ebb9
test(depth encoding): updating and cleaning video/depth encoding tests
2026-06-16 18:02:31 +02:00
f712698272
test(depth): cleaning up depth tests
2026-06-16 18:02:31 +02:00
c2416ecbcb
feat(output unit): adding support for output unit specification at dataset reading/training time
...
Co-authored-by: Wensi (Vince) Ai <59036629+wensi-ai@users.noreply.github.com >
2026-06-16 18:02:31 +02:00
6aa50cc1e5
fix(depth units): fixing depth units output for the realsense cameras
2026-06-16 18:02:31 +02:00
e17adce3ba
fix(is_depth): adding missing doctrings and is_depth arguments in video decoding functions
...
Co-authored-by: Wensi (Vince) Ai <59036629+wensi-ai@users.noreply.github.com >
2026-06-16 18:02:31 +02:00
f7010ff66c
fix(typo): fixing typo
2026-06-16 18:02:31 +02:00
f7ee453de7
fix(from_video_info): fixing early validation issue in from_video_info
2026-06-16 18:02:31 +02:00
ca7168f413
test(cleaning): cleaning up tests
2026-06-16 18:02:28 +02:00
ec6264d768
test(aggregate): extending aggregation tests to depth frames
2026-06-16 18:01:55 +02:00
d93a58a8b8
feat(tools): adding depth support in LeRobotDataset edition tools
2026-06-16 18:01:50 +02:00
92497dfcd8
feat(batched dequantization): optimizing dequantize_depth for torch based batched dequantization
2026-06-16 18:01:18 +02:00
263108d6c1
fix(TIFF): add missing quantization and cleanup for TIFF files
2026-06-16 18:01:18 +02:00
a925d20ce4
fix(typo): fixing typo
2026-06-16 18:01:18 +02:00
1f024ea3bf
fix(normalization): restricting 255 normalization to non depth/uint8 images only
2026-06-16 18:01:18 +02:00
d5f67cc7fc
fix(realsense): fixing typo in realsense serial number
2026-06-16 18:01:18 +02:00
9ab8c98494
tests(typos): fixing typos in tests
2026-06-16 18:01:18 +02:00
a561183442
fix(info): fixing info metadata update when is_depth_map was set
2026-06-16 18:01:18 +02:00
305b8d64b2
fix(pre-commit): fixing mutable defautl value
2026-06-16 18:01:18 +02:00
0a624a5cf5
feat(refactor): refactor DepthEncoderConfig quantization pipeline, so that the methods do not live in the config class. Add pixel format - channels validation.Move the default pixel format for depth in the config file.
2026-06-16 18:01:18 +02:00
d044ead377
feat(pix_fmt channels): use PyAv to check get pixel formats number of channels
2026-06-16 18:01:18 +02:00
e425fcb61a
tests(depth): adding new tests for depth integration validation
2026-06-16 18:01:17 +02:00
f08a9aea71
test(fix): fixing exisiting tests to still work with latest features
2026-06-16 18:01:17 +02:00
7d97b55cc4
chore(typos): fixing typos
2026-06-16 18:01:17 +02:00
edbd8c6f82
fix(plumbing): fixing missing parts in the depth maps pipeline
2026-06-16 18:01:17 +02:00
615954b80b
fix(stop_event): fixing stop_event race condition in camera classes
2026-06-16 18:01:17 +02:00
1c0fdfdb4b
feat(is_depth): simplifying is_depth nested name + legacy support
2026-06-16 18:01:17 +02:00
1c3ebd475f
feat(depth shape): ensuring depth maps shape is always including the channel
2026-06-16 18:01:17 +02:00
c655814788
chore(format): format code
2026-06-16 18:01:17 +02:00
a72ab14f89
feat(depth maps writer): adding support for raw depth maps recording with image writer
2026-06-16 18:01:17 +02:00
882074d707
feat(viz): render depth observations as rr.DepthImage in Viridis
2026-06-16 18:01:17 +02:00
4ae2f9f375
feat(record): plumb DepthEncoderConfig through lerobot-record
2026-06-16 18:01:17 +02:00
26099b6e03
feat(robots/so_follower): emit + populate depth keys when use_depth
2026-06-16 18:01:16 +02:00
6b395dfb24
feat(features): route 2D camera shapes to observation.depth.<key>
2026-06-16 18:01:16 +02:00
1cbabfe9a4
feat(cameras/realsense): expose async depth in metric meters
2026-06-16 18:01:16 +02:00
4744f4b913
feat(depth): wire DatasetReader to decode_depth_frames
2026-06-16 18:01:16 +02:00
9568e68b28
feat(depth): wire StreamingVideoEncoder + writer to depth encoder
2026-06-16 18:01:16 +02:00
10941c31f6
feat(depth): plumb DepthEncoderConfig through LeRobotDataset and DatasetWriter
2026-06-16 18:01:16 +02:00
a6882a048a
feat(depth): extend quantization tools to better fit the encoding/decoding pipeline
2026-06-16 18:01:16 +02:00
eb2b7d6dc3
feat(depth): persist depth metadata
2026-06-16 18:01:16 +02:00
f7f7b8c7f8
feat(video): add ffv1 to supported codecs
2026-06-16 18:01:16 +02:00
d58a324da4
feat(depth): add depth quantization helpers and tests
2026-06-16 18:01:16 +02:00
287c823f13
fix(features copy): adding deepcopy on LeRobot dataset features to avoid shallow copy leaks ( #3826 )
...
* fix(features copy): adding deepcopy on LeRobot dataset features to avoid shallow copy leaks
* tests(test): adding new test
2026-06-16 17:58:59 +02:00
58ccc01508
fix(datasets): enforce one parquet row group per episode in v3 data writes ( #3807 )
...
* fix(datasets): enforce one parquet row group per episode in v3 data writes
LeRobot v3 data shards must hold exactly one row group per episode so a
reader can fetch episode i with pq.ParquetFile(path).read_row_group(i)
(a byte-range read) instead of loading the whole shard. The recording
writer already does this (one write_table per episode); the aggregate
and lerobot-annotate re-write paths instead concatenated many episodes
and wrote them in one shot, collapsing the file to a single row group.
- io_utils: add write_table_one_row_group_per_episode (one ParquetWriter,
one write_table per episode — same pattern as the recording writer);
to_parquet_with_hf_images embeds images then writes per-episode row
groups; to_parquet_one_row_group_per_episode wraps it for plain frames
- aggregate: route non-image data writes through the per-episode writer;
leave the episodes-metadata parquet untouched (already one row/episode)
- annotate: rewrite shards via the per-episode writer instead of a single
bulk pq.write_table
- tests: invariant coverage through the aggregate (image + video) and
annotate paths
No change to on-disk schema, paths, naming, rollover thresholds, or
compression. Readers stay backward-compatible (old collapsed files load).
* Update src/lerobot/datasets/io_utils.py
Co-authored-by: Caroline Pascal <caroline8.pascal@gmail.com >
Signed-off-by: Pepijn <138571049+pkooij@users.noreply.github.com >
* Update src/lerobot/datasets/io_utils.py
Co-authored-by: Caroline Pascal <caroline8.pascal@gmail.com >
Signed-off-by: Pepijn <138571049+pkooij@users.noreply.github.com >
* fix(datasets): correct indentation and add strict= in row-group helper
The web-edited numpy version of write_table_one_row_group_per_episode had an
over-indented line (IndentationError, breaking pre-commit + test collection)
and a zip() without strict=. Fix both; behaviour unchanged.
---------
Signed-off-by: Pepijn <138571049+pkooij@users.noreply.github.com >
Co-authored-by: Caroline Pascal <caroline8.pascal@gmail.com >
2026-06-16 12:15:48 +02:00
CarolinePascal