Add files using upload-large-folder tool

README.md

@@ -0,0 +1,137 @@
+---
+license: cc-by-4.0
+pretty_name: V33DA
+task_categories:
+- audio-classification
+- video-classification
+tags:
+- animal-behavior
+- bioacoustics
+- bird-vocalization
+- multimodal
+- 3d-pose
+- localization
+- attribution
+- radio-telemetry
+- accelerometer
+size_categories:
+- 10K<n<100K
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: v33da-*.parquet
+---
+
+# V33DA: Benchmarking Vocal Attribution in Freely-Behaving Zebra Finches
+
+V33DA is a multimodal benchmark for vocal attribution and 3D vocal localization in small groups of freely behaving zebra finches recorded in the BirdPark aviary. Each released sample is a single vocalization event with synchronized microphone audio, video, 3D pose, radio telemetry, and the accelerometer-derived on-body verification signal used to establish the ground-truth caller label.
+
+The dataset is designed so that caller labels do not reduce to the same room-acoustic attribution problem solved by the benchmark models. Instead, the release follows the paper’s extraction pipeline: candidate events are proposed from the body-mounted verification channel, manually reviewed, and then filtered by geometric and motion consistency before inclusion.
+
+## Paper Summary
+
+Recordings were collected from groups of 3--4 zebra finches during natural social interactions. The sensing setup aligns the same event across:
+
+- five cage microphones sampled at 24,414 Hz
+- three synchronized camera views
+- five 3D body keypoints per bird
+- 21 per-bird radio telemetry signals per frame
+- the raw accelerometer-derived vibration signal carried by each bird
+
+The benchmark supports a task introduced in the paper:
+
+1. vocal attribution: identify which visible bird produced the vocalization
+
+## Modalities
+
+Each parquet row corresponds to one released vocalization event and includes:
+
+- `audio_path`: relative path to the multichannel cage-microphone WAV
+- `accelerometer_path`: relative path to the multichannel accelerometer WAV
+- `video_path`: relative path to the aligned composite MP4 clip
+- `keypoints_3d`: triangulated 3D body pose
+- `keypoints_2d_top`: 2D keypoints in the top-view image space
+- `keypoints_2d_back`: 2D keypoints in the back-view image space
+- `radio_*`: synchronized per-bird radio telemetry arrays
+- `bird_color`: color identity used as the source of truth for caller assignment
+- `vocalizer_idx`: index of the vocalizing bird
+
+Media is stored as external files and referenced from parquet by path.
+
+## Labeling and Filtering
+
+The paper’s released set is built from a reviewed candidate pool. After manual verification, samples are retained only if they satisfy the release filters:
+
+- overlapping vocalizations removed
+- 3D-to-2D reprojection error `<= 40 px`
+- maximum frame-to-frame 2D displacement `<= 40 px`
+- no NaNs in released 3D or 2D keypoints
+
+Color is the source of truth for the released vocalizer identity.
+
+## Release Contents
+
+This standalone release contains:
+
+- `v33da-*.parquet`: pooled parquet shards for the full release
+- `audio/`: multichannel microphone WAV files
+- `accelerometer/`: multichannel accelerometer WAV files
+- `clips/`: aligned composite MP4 clips
+- `calibrations/`: per-experiment camera calibration files
+- `explore.ipynb`: notebook for dataset exploration and reprojection
+- `metadata.json`: release schema and filter metadata
+
+## Dataset Size
+
+- released samples: `33,625`
+- pooled parquet shards: `23`
+- experiments: `3`
+- individuals: `11`
+
+Per-group counts in this standalone release:
+
+- `juvExpBP01 / blue`: `5,735`
+- `juvExpBP01 / peach`: `2,277`
+- `juvExpBP01 / red`: `3,963`
+- `juvExpBP01 / white`: `4,244`
+- `juvExpBP02 / blue`: `4,923`
+- `juvExpBP02 / peach`: `4,589`
+- `juvExpBP02 / red`: `2,065`
+- `juvExpBP05 / brown`: `3,633`
+- `juvExpBP05 / purple`: `2,140`
+- `juvExpBP05 / yellow`: `56`
+
+## Using the Dataset
+
+The canonical release format is the pooled `v33da-*.parquet` set. Each binary array column is serialized with `numpy.save`, so decoding is:
+
+```python
+import io
+import numpy as np
+
+arr = np.load(io.BytesIO(raw_bytes))
+```
+
+The included `explore.ipynb` notebook shows how to:
+
+- load parquet shards
+- inspect audio, accelerometer, video, pose, and radio
+- use the shipped camera calibrations to reproject 3D keypoints into 2D views
+
+## Limitations
+
+As discussed in the paper, V33DA is collected in one aviary with one recording geometry across three experiments. Cross-experiment evaluation therefore combines bird-identity shift, group-composition shift, and recording-date shift. The released benchmark also excludes overlapping vocalizations and filters out samples with poor geometric consistency or excessive motion.
+
+## Citation
+
+If you use this dataset, please cite the accompanying paper:
+
+```bibtex
+@inproceedings{basha2026v33da,
+  title={V33DA: Benchmarking Vocal Attribution in Freely-Behaving Zebra Finches},
+  author={Basha, Maris and Wang, Yuhang and Chen, Xiaoran and Cheng, Longbiao and Yapura, Luca and Salzmann, Mathieu and Hahnloser, Richard},
+  booktitle={XXX},
+  year={2026}
+}
+```

accelerometer/juvExpBP01/BP_2021-07-31_06-21-00_336025_0060000_1418.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5acba9144c37aac5e75dff6f05d5e19a94616e31c33ac3a2b8d399aaf0161048
+size 37608

accelerometer/juvExpBP01/BP_2021-07-31_13-27-26_336025_1280000_5527.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad221ea097c18d62ef37eb1e954b938a1033f44a2a8e7be8143ecbb7b5d88455
+size 35256

accelerometer/juvExpBP01/BP_2021-07-31_19-44-55_336025_2360000_16572.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c343e3f5e8b5d1ae797ff4cd0c3e00ea3003c7d688d62ad763af7abb8d73c280
+size 58312

accelerometer/juvExpBP01/BP_2021-07-31_19-44-55_336025_2360000_3382.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90bb47ff7c5b37c6477d6194d6d6b6bc0967708b81a3b99774bc3ea8d8e6f9f9
+size 43080

accelerometer/juvExpBP01/BP_2021-08-01_08-40-49_798997_0460000_1351.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4bd42eb2abdf5fcf48f7ef43b5b7fd8373b10ac0f8ea228348ec741887c57e7
+size 41512

accelerometer/juvExpBP01/BP_2021-08-01_08-40-49_798997_0460000_1750.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7554a44ff9cf93d11fb8230bab0fb5aa6b9622e3c2618f4a2b8e4e021998b150
+size 68856

accelerometer/juvExpBP01/BP_2021-08-01_09-43-44_798997_0640000_10522.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d95ac2010c317b0b428ab50502e2979b436a67484a559dc4899e1b9511a0afd4
+size 53224

accelerometer/juvExpBP01/BP_2021-08-01_09-43-44_798997_0640000_3269.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:642deb1487880047843b9605930dba43e1a13f0616aa42f2175787bc74fa5586
+size 54792

accelerometer/juvExpBP01/BP_2021-08-01_09-43-44_798997_0640000_6569.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c95a4edb11342c6406910a9d40da787d2f21bc1c1368c4bdc9e5afc85fb21c96
+size 41896

accelerometer/juvExpBP01/BP_2021-08-01_10-25-40_798997_0760000_15375.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42d92ca654a4ebbf9e512ec21592da4ef578862bf4e1303fa77c281f20112a95
+size 107144

accelerometer/juvExpBP01/BP_2021-08-03_06-41-59_796988_0120000_35.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e41b98599ed73f1690335b218db26744e3c43217c75643b3de7a2183deacf210
+size 58312

accelerometer/juvExpBP01/BP_2021-08-03_08-47-48_796988_0480000_732.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:766eb1bcb78550b9161fb3b36af0f3d70b26e476e6391aae5908d486b389979a
+size 46200

accelerometer/juvExpBP01/BP_2021-08-03_14-23-21_796988_1440000_16629.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:266224868b9c9f82c13db292d881ed911d1b3c609e6d80c847a0670a3a84b9bf
+size 134472

accelerometer/juvExpBP01/BP_2021-08-03_15-12-17_796988_1580000_154.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6f3703d1d8be07b83e791f671b26874a993ea59a4975a6486d1dfeda3dbc501
+size 44632

accelerometer/juvExpBP01/BP_2021-08-03_15-26-16_796988_1620000_19405.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa777a609d99259d0b6750dbba2e6e2aa10008d5a93c56f2f62348b05b6e1bb9
+size 46600

accelerometer/juvExpBP01/BP_2021-08-03_15-26-16_796988_1620000_19479.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:240457a3e41154d43634819bd0fafe246d9072cda95aa83b793c2b6e18a72fbb
+size 46984

accelerometer/juvExpBP01/BP_2021-08-04_08-19-51_714275_0400000_19599.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29d115c4fd9c16ec5a3619bd90d66c8bdaa663a451c787980247c9e44407df08
+size 39560

accelerometer/juvExpBP01/BP_2021-08-04_08-40-49_714275_0460000_2256.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1b7f55817b1aeb8d5f96669113ebef03e885517ea03af3ceb6e017a7d7d4b0a
+size 117304

accelerometer/juvExpBP01/BP_2021-08-04_17-11-07_714275_1920000_51.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91ac45af7c39642325074f43dbdeece9a15fea448a2571fc00090c575bf04945
+size 51272

accelerometer/juvExpBP01/BP_2021-08-04_18-14-02_714275_2100000_9749.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef13e36e6085fba6022f367f8389ba146939c0ab30582123f175126dc57ea83d
+size 102840

accelerometer/juvExpBP01/BP_2021-08-05_06-00-02_416952_0000000_13046.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02d4f474d42668bc3b598caa65d536fe19aae28ba6a3940caf260f6284956334
+size 48920

accelerometer/juvExpBP01/BP_2021-08-05_06-28-00_416952_0080000_2309.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93797d523f7fe1a9959601353bcae59529c11cb453afad93d61ce714ef59173e
+size 41896

accelerometer/juvExpBP01/BP_2021-08-05_10-18-41_416952_0740000_12886.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:357d7e6e48477ba06c24c2ff7798ca8af3a9e2d0f74db9268ead49dfe51e5659
+size 43080

accelerometer/juvExpBP01/BP_2021-08-05_10-18-41_416952_0740000_18354.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07298000bd1c2a514a8ecf649e94f451a52238bea80476df7c7acb5ba5de03b1
+size 34088

accelerometer/juvExpBP01/BP_2021-08-05_10-18-41_416952_0740000_7230.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:022912d1bcb791d2e1e9a90b9dfc541fe0f7a744acffe6c01349f2820cedea94
+size 89160

accelerometer/juvExpBP01/BP_2021-08-05_10-18-41_416952_0740000_7251.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddd3139a349e2a937309e583756d75aafd724c7ee44967966526a475f5e7d172
+size 103624

accelerometer/juvExpBP01/BP_2021-08-05_10-32-40_416952_0780000_19292.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be3a5a3414c5ef1f0e5d59d7651f865c01e89b562bfe80174ad55368c3ec4c7e
+size 33704

accelerometer/juvExpBP01/BP_2021-08-05_11-00-37_416952_0860000_636.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9da4fe854d20c3628fd630df75d191c9bd6fa884500ca9189413c51114e163dc
+size 43080

accelerometer/juvExpBP01/BP_2021-08-06_08-54-48_024871_0500000_9360.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4907c07c60fe89d5e6e61368240f4b035e1bc403ada77e34ef51d793e5fc3600
+size 41896

accelerometer/juvExpBP01/BP_2021-08-06_09-57-43_024871_0680000_1527.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e4c62db87f3bac0d0ccededfb62126cab21ee77e31077fd2eb5aed2b6ea0f76
+size 45016

accelerometer/juvExpBP01/BP_2021-08-06_11-14-37_024871_0900000_5486.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d918f340e9abb3991976b7b0b11a424c783cf0125477ccfdd0b55de3f0fb846
+size 31752

accelerometer/juvExpBP01/BP_2021-08-06_13-27-26_024871_1280000_9898.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b607d949c389d4170fc6146b70f48024004eba1864e70b2ca6b6f2dfaa2b3c6
+size 33704

accelerometer/juvExpBP01/BP_2021-08-06_19-37-56_024871_2340000_758.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82696c51760eb0441f9b560d54f2c02745e74f09e094212641e43fd718f70668
+size 55576

accelerometer/juvExpBP01/BP_2021-08-07_07-09-56_584350_0200000_616.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:887a6f2598b32d75e42c8e586165522a8d8686a81814ea1816437379b7e48093
+size 45032

accelerometer/juvExpBP01/BP_2021-08-07_09-22-45_584350_0580000_681.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95b72556549146155d4afbf6bba285c236953462b31eeec947022adc4c3e85a7
+size 50888

accelerometer/juvExpBP01/BP_2021-08-07_19-44-55_584350_2360000_2122.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be2204adfe96cac4d5d404f22a7ffb9f0e47155a81ecdaf76aace0504ebdd279
+size 58312

accelerometer/juvExpBP01/BP_2021-08-08_08-05-52_377038_0360000_10512.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d39f560a4198379d44079c7d20f3b44f29f4cfd0dd93145c61ad9fe0b93de31
+size 179800

accelerometer/juvExpBP01/BP_2021-08-08_08-05-52_377038_0360000_7610.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9ed60b44dc4bfcda61da2d6afcd925d53613bdb3bbe9fb177bbcd64b0a51be5
+size 54792

accelerometer/juvExpBP01/BP_2021-08-08_12-03-33_377038_1040000_241.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdc8bbd639c0f378c3a0095b86ea72a13f4eb15667159044f6cb2bca8272df54
+size 32904

accelerometer/juvExpBP01/BP_2021-08-08_12-38-30_377038_1140000_81.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a5482dc90e0a6c1603eeba6b690e460bb152947289038dc557d60bcebafbff8
+size 54408

accelerometer/juvExpBP01/BP_2021-08-08_12-38-30_377038_1140000_9110.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29aa00705e73e74c7bc6ee5766bae12800ebf8f605bb543dc65a02e217cc2ba6
+size 65736

accelerometer/juvExpBP01/BP_2021-08-08_19-44-55_377038_2360000_13875.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a217d6a9aab45e13923d99415dd61b85b4a84c0ff46c2cc3fb08d04bd2c51aa5
+size 48536

accelerometer/juvExpBP01/BP_2021-08-09_10-11-42_504687_0720000_14812.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f7c044264f3248044b68d0f04c6304237b411665003a0be29692a436d778912
+size 35256

calibrations/.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

calibrations/README.md

@@ -0,0 +1,38 @@
+---
+license: mit
+tags:
+  - zebra-finch
+  - camera-calibration
+  - birdpark
+---
+
+# BirdPark Camera Calibration & Metadata
+
+Camera intrinsics, extrinsics, and bird-color metadata for the BirdPark multi-view tracking setup (3 cameras: top, back, side).
+
+## Download
+
+```bash
+pip install huggingface_hub
+```
+
+```python
+from huggingface_hub import snapshot_download
+snapshot_download("songbirdini/birdpark_calibration", local_dir="tracking/data")
+```
+
+Or from the TCBI analysis project:
+
+```bash
+python -m tracking.download_calibration
+```
+
+## Contents
+
+| File | Description |
+|------|-------------|
+| `calibration_{view}.npz` | Camera intrinsic matrix (K) + distortion coefficients |
+| `camera_pose_{view}.pkl` | Camera extrinsic rotation (rvec) + translation (tvec) |
+| `metadata.csv` | Experiment -> bird ID -> backpack color mapping |
+
+Views: `top` (1292x1292), `back` (2202x724), `side` (774x1292, rotated 90 degrees)

calibrations/reference_points.csv

@@ -0,0 +1,52 @@
+Unnamed: 0,X（mm）,Y（mm）,Z（mm）,Index,position
+116.0,323.8442214,665.019271042,1081.815583062,1,wedge_wall
+116.0,348.072516622,665.019271042,1081.815583062,2,wedge_wall
+116.0,323.8442214,665.019271042,1106.043878284,3,wedge_wall
+116.0,348.072516622,665.019271042,1106.043878284,4,wedge_wall
+116.0,318.072516622,640.019271042,1106.043878284,5,wedge_wall
+116.0,348.072516622,640.019271042,1106.043878284,6,wedge_wall
+116.0,348.072516622,640.019271042,1076.043878284,7,wedge_wall
+116.0,1041.351578025,665.019271042,1425.482844503,8,wedege_door
+116.0,1017.123282803,665.019271042,1449.711139724,9,wedege_door
+116.0,1041.351578025,665.019271042,1449.711139724,10,wedege_door
+116.0,1017.123282803,640.019271042,1425.482844503,11,wedege_door
+116.0,1047.123282803,640.019271042,1425.482844503,12,wedege_door
+116.0,1017.123282803,640.019271042,1455.482844503,13,wedege_door
+116.0,1046.816821976,511.40242249,1046.34693913,14,nest_wall
+116.0,1046.816821976,511.40242249,1076.34693913,15,nest_wall
+116.0,1046.816821976,511.40242249,1142.444265892,16,nest_wall
+116.0,1046.816821976,511.40242249,1172.444265892,17,nest_wall
+116.0,1076.816821976,511.40242249,1172.444265892,18,nest_wall
+116.0,1046.866821976,536.40242249,1172.89693913,19,nest_wall
+116.0,1046.866821976,633.40242249,1174.562380619,20,nest_wall
+116.0,323.619270346,664.044900977,1454.04025791,21,nest_door
+116.0,418.374120452,664.044900977,1449.93949076,22,nest_door
+116.0,288.378977449,511.40242249,1455.179783657,23,nest_door
+116.0,318.378977449,511.40242249,1455.179783657,24,nest_door
+116.0,318.378977449,536.40242249,1455.179783657,25,nest_door
+116.0,414.928977449,536.40242249,1455.229783657,26,nest_door
+116.0,384.474745717,511.40242249,1455.179783657,27,nest_door
+116.0,414.474745717,511.40242249,1455.179783657,28,nest_door
+116.0,414.474745717,511.40242249,1485.179783657,29,nest_door
+116.0,241.599015305,253.988072442,999.741100871,30,plate_left
+116.0,678.599014328,255.488072442,999.770322483,31,plate_left
+116.0,241.563441168,255.488072442,1531.741099682,32,plate_left
+116.0,678.56344019,253.988072442,1531.770321294,33,plate_left
+116.0,686.40473411,255.488072442,999.770407744,34,plate_right
+116.0,1123.40473411,255.488072442,999.770407744,35,plate_right
+116.0,686.40473411,255.488072442,1531.770407744,36,plate_right
+116.0,1123.40473411,255.488072442,1531.770407744,37,plate_right
+116.0,1017.123282803,665.019271042,1425.482844503,38,wedege_door
+116.0,-90.761717388,1529.988072442,667.932180663,A,box
+116.0,-90.761717388,1529.988072442,1563.932180663,D,box
+116.0,-90.761717388,235.988072442,1563.932180663,C,box
+116.0,-90.761717388,235.988072442,667.932180663,B,box
+116.0,1155.238282612,1529.988072442,667.932180663,E,box
+116.0,1155.238282612,1529.988072442,1563.932180663,H,box
+116.0,1155.238282612,235.988072442,667.932180663,F,box
+116.0,1155.238282612,235.988072442,1563.932180663,G,box
+,855.0,1265.0,1266.0,,Mic1(roughly)
+,1020.0,775.0,1266.0,,Mic2-right
+,303.0,775.0,1266.0,,Mic3-left
+,682.0,775.0,1060.0,,Mic4-back
+,682.0,775.0,1434.0,,Mic5-door

explore.ipynb

@@ -0,0 +1,396 @@
+{
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "# v33da Dataset Explorer\n",
+        "\n",
+        "Standalone exploration notebook for `v33da`.\n",
+        "\n",
+        "This notebook mirrors the old `v33dl` dataset explorer style, but targets the standalone `v33da` package directly. It walks through:\n",
+        "\n",
+        "1. Loading pooled parquet shards\n",
+        "2. Picking a sample\n",
+        "3. Audio and accelerometer signals\n",
+        "4. Video clips\n",
+        "5. 3D pose and 2D keypoints\n",
+        "6. Camera calibration and reprojection\n",
+        "7. Radio telemetry"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 0. Setup\n",
+        "\n",
+        "If you already have the standalone dataset locally, just run the next cell. The notebook resolves the dataset root automatically."
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from __future__ import annotations\n",
+        "\n",
+        "import io\n",
+        "import pickle\n",
+        "from pathlib import Path\n",
+        "\n",
+        "import cv2\n",
+        "import matplotlib.pyplot as plt\n",
+        "import numpy as np\n",
+        "import pyarrow.parquet as pq\n",
+        "import soundfile as sf\n",
+        "\n",
+        "R_ZUP = np.array([[1, 0, 0], [0, 0, -1], [0, 1, 0]], dtype=np.float64)\n",
+        "KEYPOINT_NAMES = [\"beak\", \"head\", \"backpack\", \"tailbe\", \"tailend\"]\n",
+        "RADIO_KEYS = [\n",
+        "    \"frq\", \"thetaS\", \"thetaA\", \"thetaB\", \"thetaC\", \"thetaD\",\n",
+        "    \"phiA\", \"phiB\", \"phiC\", \"phiD\", \"phiM\",\n",
+        "    \"powRA\", \"powRB\", \"powRC\", \"powRD\", \"powRM\",\n",
+        "    \"powNA\", \"powNB\", \"powNC\", \"powND\", \"powNM\",\n",
+        "]\n",
+        "\n",
+        "def resolve_dataset_root() -> Path:\n",
+        "    cwd = Path.cwd()\n",
+        "    if (cwd / 'metadata.json').exists() and (cwd / 'audio').exists():\n",
+        "        return cwd\n",
+        "    candidate = Path('/home/songbird/code/v33da/data/v33da')\n",
+        "    if candidate.exists():\n",
+        "        return candidate\n",
+        "    raise FileNotFoundError('Could not locate the standalone v33da dataset root.')\n",
+        "\n",
+        "DATASET_ROOT = resolve_dataset_root()\n",
+        "DATASET_ROOT"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 1. Load & Inspect the Pooled Parquet Shards"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "pooled_shards = sorted(DATASET_ROOT.glob('v33da-*.parquet'))\n",
+        "len(pooled_shards), pooled_shards[:3]"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "first_table = pq.read_table(pooled_shards[0])\n",
+        "first_table.schema"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 2. Pick a Sample"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "def decode_array(blob: bytes) -> np.ndarray:\n",
+        "    return np.load(io.BytesIO(blob))\n",
+        "\n",
+        "def resolve_media_path(path_str: str) -> Path:\n",
+        "    path = Path(path_str)\n",
+        "    if path.is_absolute():\n",
+        "        return path\n",
+        "    return DATASET_ROOT / path\n",
+        "\n",
+        "def load_sample(shard_index: int = 0, row_index: int = 0):\n",
+        "    row = pq.read_table(pooled_shards[shard_index]).slice(row_index, 1).to_pylist()[0]\n",
+        "    row['keypoints_3d'] = decode_array(row['keypoints_3d'])\n",
+        "    row['keypoints_2d_top'] = decode_array(row['keypoints_2d_top'])\n",
+        "    row['keypoints_2d_back'] = decode_array(row['keypoints_2d_back'])\n",
+        "    radio = {}\n",
+        "    for key in RADIO_KEYS:\n",
+        "        radio[key] = decode_array(row[f'radio_{key}'])\n",
+        "    row['radio'] = radio\n",
+        "    row['audio_file'] = resolve_media_path(row['audio_path'])\n",
+        "    row['accelerometer_file'] = resolve_media_path(row['accelerometer_path'])\n",
+        "    row['video_file'] = resolve_media_path(row['video_path'])\n",
+        "    return row\n",
+        "\n",
+        "sample = load_sample(shard_index=0, row_index=0)\n",
+        "sample['id'], sample['experiment'], sample['bird_color'], sample['video_file'].name"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 3. Audio and Accelerometer"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "audio, sr_audio = sf.read(sample['audio_file'])\n",
+        "accel, sr_accel = sf.read(sample['accelerometer_file'])\n",
+        "audio.shape, sr_audio, accel.shape, sr_accel"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "fig, axes = plt.subplots(2, 1, figsize=(14, 6), sharex=False)\n",
+        "axes[0].plot(audio)\n",
+        "axes[0].set_title('Microphone audio (all channels)')\n",
+        "axes[1].plot(accel)\n",
+        "axes[1].set_title('Accelerometer audio (all channels)')\n",
+        "plt.tight_layout()"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 4. Video"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "cap = cv2.VideoCapture(str(sample['video_file']))\n",
+        "ok, frame_bgr = cap.read()\n",
+        "cap.release()\n",
+        "frame = cv2.cvtColor(frame_bgr, cv2.COLOR_BGR2RGB) if ok else None\n",
+        "frame.shape if frame is not None else None"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "plt.figure(figsize=(12, 8))\n",
+        "plt.imshow(frame)\n",
+        "plt.title(sample['video_file'].name)\n",
+        "plt.axis('off')"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 5. 3D Pose and 2D Keypoints"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "kp3d = sample['keypoints_3d']\n",
+        "kp2d_top = sample['keypoints_2d_top']\n",
+        "kp2d_back = sample['keypoints_2d_back']\n",
+        "kp3d.shape, kp2d_top.shape, kp2d_back.shape"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "mid = kp3d.shape[0] // 2\n",
+        "mid_beaks = kp3d[mid, :, 0]\n",
+        "\n",
+        "fig = plt.figure(figsize=(7, 6))\n",
+        "ax = fig.add_subplot(111, projection='3d')\n",
+        "ax.scatter(mid_beaks[:, 0], mid_beaks[:, 1], mid_beaks[:, 2], s=60)\n",
+        "for bird_idx, pt in enumerate(mid_beaks):\n",
+        "    ax.text(pt[0], pt[1], pt[2], str(bird_idx))\n",
+        "ax.set_xlabel('X (mm)')\n",
+        "ax.set_ylabel('Y (mm)')\n",
+        "ax.set_zlabel('Z (mm)')\n",
+        "ax.set_title('Mid-frame beak positions')"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 6. Camera Calibration and Reprojection"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "def load_calibration(experiment: str, view: str):\n",
+        "    exp_dir = DATASET_ROOT / 'calibrations' / experiment\n",
+        "    npz = np.load(exp_dir / f'calibration_{view}.npz')\n",
+        "    with open(exp_dir / f'camera_pose_{view}.pkl', 'rb') as handle:\n",
+        "        pose = pickle.load(handle)\n",
+        "    return {\n",
+        "        'K': npz['mtx'],\n",
+        "        'dist': npz['dist'],\n",
+        "        'rvec': pose['rvec'],\n",
+        "        'tvec': pose['tvec'],\n",
+        "    }\n",
+        "\n",
+        "def project_zup(points_xyz_mm: np.ndarray, calibration: dict) -> np.ndarray:\n",
+        "    pts = np.asarray(points_xyz_mm, dtype=np.float64)\n",
+        "    pts_cam = (R_ZUP @ pts.T).T\n",
+        "    proj, _ = cv2.projectPoints(pts_cam, calibration['rvec'], calibration['tvec'], calibration['K'], calibration['dist'])\n",
+        "    return proj.reshape(-1, 2)\n",
+        "\n",
+        "def reproject_view(row, view: str):\n",
+        "    cal = load_calibration(row['experiment'], view)\n",
+        "    kp3d_mid = row['keypoints_3d'][mid]\n",
+        "    kp2d_mid = row[f'keypoints_2d_{view}'][mid]\n",
+        "    projected = np.full_like(kp2d_mid, np.nan, dtype=np.float64)\n",
+        "    errors = np.full(kp2d_mid.shape[:2], np.nan, dtype=np.float64)\n",
+        "    for bird_idx in range(kp3d_mid.shape[0]):\n",
+        "        valid3d = ~np.any(np.isnan(kp3d_mid[bird_idx]), axis=1)\n",
+        "        valid2d = ~np.any(np.isnan(kp2d_mid[bird_idx]), axis=1)\n",
+        "        if valid3d.any():\n",
+        "            projected[bird_idx, valid3d] = project_zup(kp3d_mid[bird_idx, valid3d], cal)\n",
+        "        valid = valid2d & valid3d\n",
+        "        if valid.any():\n",
+        "            errors[bird_idx, valid] = np.linalg.norm(projected[bird_idx, valid] - kp2d_mid[bird_idx, valid], axis=1)\n",
+        "    return kp2d_mid, projected, errors\n",
+        "\n",
+        "top_obs, top_proj, top_err = reproject_view(sample, 'top')\n",
+        "back_obs, back_proj, back_err = reproject_view(sample, 'back')\n",
+        "np.nanmedian(top_err), np.nanmedian(back_err)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "def plot_reprojection(ax, observed, projected, title):\n",
+        "    colors = ['tab:red', 'tab:orange', 'tab:blue', 'tab:green']\n",
+        "    for bird_idx in range(observed.shape[0]):\n",
+        "        obs = observed[bird_idx]\n",
+        "        proj = projected[bird_idx]\n",
+        "        valid_obs = ~np.any(np.isnan(obs), axis=1)\n",
+        "        valid_proj = ~np.any(np.isnan(proj), axis=1)\n",
+        "        if valid_obs.any():\n",
+        "            ax.scatter(obs[valid_obs, 0], obs[valid_obs, 1], c=colors[bird_idx % len(colors)], marker='o', alpha=0.8)\n",
+        "        if valid_proj.any():\n",
+        "            ax.scatter(proj[valid_proj, 0], proj[valid_proj, 1], c=colors[bird_idx % len(colors)], marker='x', alpha=0.8)\n",
+        "        valid = valid_obs & valid_proj\n",
+        "        for kp_idx in np.where(valid)[0]:\n",
+        "            ax.plot([obs[kp_idx, 0], proj[kp_idx, 0]], [obs[kp_idx, 1], proj[kp_idx, 1]], color=colors[bird_idx % len(colors)], alpha=0.35)\n",
+        "    ax.set_title(title)\n",
+        "    ax.invert_yaxis()\n",
+        "    ax.set_aspect('equal')\n",
+        "    ax.grid(alpha=0.2)\n",
+        "\n",
+        "fig, axes = plt.subplots(1, 2, figsize=(14, 6))\n",
+        "plot_reprojection(axes[0], top_obs, top_proj, f'top | median reproj = {np.nanmedian(top_err):.2f}px')\n",
+        "plot_reprojection(axes[1], back_obs, back_proj, f'back | median reproj = {np.nanmedian(back_err):.2f}px')\n",
+        "plt.tight_layout()"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 7. Radio Telemetry"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "sample['radio']['powRA'].shape, sample['radio']['thetaS'].shape"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "fig, axes = plt.subplots(2, 1, figsize=(14, 6), sharex=True)\n",
+        "axes[0].plot(sample['radio']['powRA'].T)\n",
+        "axes[0].set_title('radio powRA per bird')\n",
+        "axes[1].plot(sample['radio']['thetaS'].T)\n",
+        "axes[1].set_title('radio thetaS per bird')\n",
+        "plt.tight_layout()"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "## 8. Iterate Efficiently\n",
+        "\n",
+        "Use this pattern if you want to scan pooled shards without loading everything into RAM."
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "for shard in pooled_shards[:2]:\n",
+        "    table = pq.read_table(shard, columns=['experiment', 'bird_color'])\n",
+        "    print(shard.name, table.num_rows)"
+      ]
+    }
+  ],
+  "metadata": {
+    "kernelspec": {
+      "display_name": "Python 3",
+      "language": "python",
+      "name": "python3"
+    },
+    "language_info": {
+      "codemirror_mode": {
+        "name": "ipython",
+        "version": 3
+      },
+      "file_extension": ".py",
+      "mimetype": "text/x-python",
+      "name": "python",
+      "nbconvert_exporter": "python",
+      "pygments_lexer": "ipython3",
+      "version": "3.12"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 5
+}

metadata.json

@@ -0,0 +1,78 @@
+{
+  "dataset_name": "v33da",
+  "per_bird_limit": null,
+  "num_rows": 33625,
+  "groups": {
+    "juvExpBP01/blue": 5735,
+    "juvExpBP01/peach": 2277,
+    "juvExpBP01/red": 3963,
+    "juvExpBP01/white": 4244,
+    "juvExpBP02/blue": 4923,
+    "juvExpBP02/peach": 4589,
+    "juvExpBP02/red": 2065,
+    "juvExpBP05/brown": 3633,
+    "juvExpBP05/purple": 2140,
+    "juvExpBP05/yellow": 56
+  },
+  "schema": [
+    "id",
+    "experiment",
+    "date",
+    "video_stem",
+    "bird_name",
+    "bird_color",
+    "vocalizer_idx",
+    "n_frames",
+    "keypoints_3d",
+    "keypoints_2d_top",
+    "keypoints_2d_back",
+    "audio",
+    "audio_path",
+    "accelerometer",
+    "accelerometer_path",
+    "video_path",
+    "radio_frq",
+    "radio_thetaS",
+    "radio_thetaA",
+    "radio_thetaB",
+    "radio_thetaC",
+    "radio_thetaD",
+    "radio_phiA",
+    "radio_phiB",
+    "radio_phiC",
+    "radio_phiD",
+    "radio_phiM",
+    "radio_powRA",
+    "radio_powRB",
+    "radio_powRC",
+    "radio_powRD",
+    "radio_powRM",
+    "radio_powNA",
+    "radio_powNB",
+    "radio_powNC",
+    "radio_powND",
+    "radio_powNM"
+  ],
+  "filters": {
+    "max_reproj_error_px": 40.0,
+    "max_global_displacement_px": null,
+    "max_intra_frame_displacement_px": 40.0,
+    "motion_filter_mode": "frame_to_frame_only",
+    "nan_policy": "drop_rows_with_any_nan_in_3d_or_2d",
+    "overlap_pad_sec": 0.01
+  },
+  "overlap_removed_before_build": 3784,
+  "frames": {
+    "description": "Full composite video chunks aligned to each vocalization window",
+    "format": "MP4 (mp4v codec)",
+    "frame_rate_hz": 47.6837158203125,
+    "pad_frames": 1
+  },
+  "external_media": {
+    "audio_rows_with_paths": 33625,
+    "accelerometer_rows_with_paths": 33625,
+    "video_rows_with_paths": 33625,
+    "media_jobs_path": "media_jobs.jsonl",
+    "media_results_path": "media_results.jsonl"
+  }
+}

split_summary.json

@@ -0,0 +1,44 @@
+{
+  "dataset_root": "/home/songbird/code/v33da/data/v33da",
+  "total_rows": 33625,
+  "session_split_counts": {
+    "test": 7391,
+    "train": 19365,
+    "val": 6869
+  },
+  "generalization_split_counts": {
+    "test": 5829,
+    "train": 21491,
+    "val": 6305
+  },
+  "session_by_experiment": {
+    "train": {
+      "juvExpBP01": 12881,
+      "juvExpBP02": 4455,
+      "juvExpBP05": 2029
+    },
+    "val": {
+      "juvExpBP01": 1008,
+      "juvExpBP02": 4155,
+      "juvExpBP05": 1706
+    },
+    "test": {
+      "juvExpBP01": 2330,
+      "juvExpBP02": 2967,
+      "juvExpBP05": 2094
+    }
+  },
+  "generalization_by_experiment": {
+    "train": {
+      "juvExpBP01": 12881,
+      "juvExpBP02": 8610
+    },
+    "val": {
+      "juvExpBP01": 3338,
+      "juvExpBP02": 2967
+    },
+    "test": {
+      "juvExpBP05": 5829
+    }
+  }
+}