README.md

---
license: cc-by-nc-4.0
task_categories:
- robotics
- image-to-3d
language:
- en
tags:
- vitra
- mano
- hand-object-interaction
- egocentric
---

# ssv2-annotations-v1

**Annotations only — no images, no video.**

VITRA-style hand episodes for **Something-Something V2**, with per-hand instructions and paraphrases.

| | |
|---|---|
| episodes | **52,706** |
| training samples (`index_frame_pair` rows) | **1,124,722** |
| annotation | MANO pose + world/camera joints + per-frame extrinsics |
| text | one instruction per episode + 2.09 paraphrases on average |
| **images / video** | **not included** — see *Getting the frames* below |

### What we did

**Episodes are NOT ours** — they are VITRA-1M's official segmentation, used unchanged.

**Instructions are ours.** Two rounds, both with `Qwen3.5-122B-A10B-FP8`:
round 1 captions 8 frames per episode with the palm's future trajectory drawn on them;
round 2 checks the sentence belongs to that hand, strips same-hand references
("Rinse the right hand." -> "Rinse the hand.", because training already prepends
`Left hand: ... Right hand: ...`), and writes 1-3 paraphrases.


### Files

```
ssv2.tar                 ->  Annotation/ssv2/episodic_annotations/*.npy
episode_frame_index.npz        index_frame_pair (N,2) uint32 + index_to_episode_id (E,)
```

`index_frame_pair` **row number is the sample id**: row r = (episode ordinal, frame within
that episode). `len(index_frame_pair)` is the size of the training set.

```python
import numpy as np, tarfile
# tar -xf ssv2.tar
z = np.load("episode_frame_index.npz", allow_pickle=True)
ep_slot, frame_id = z["index_frame_pair"][sample_id]
eid = str(z["index_to_episode_id"][ep_slot])
d = np.load(f"Annotation/ssv2/episodic_annotations/{eid}.npy", allow_pickle=True).item()
rgb_frame_id = int(d["video_decode_frame"][frame_id])
```

Each `.npy` is a dict with `video_name`, `video_decode_frame`, `intrinsics`,
per-frame `extrinsics` (world->camera), `anno_type` (which hand this episode is for),
`text`, `text_rephrase`, and a `left`/`right` dict holding `beta`, `hand_pose`,
`global_orient_worldspace`, `transl_worldspace`, `joints_worldspace`, `kept_frames`.
`text[hand] = [(sentence, (0, T))]` and `text_rephrase[hand] = [([paraphrases...], (0, T))]`.

### Getting the frames

`video_decode_frame` indexes the **source** video, which we do not redistribute.
Get it from Something-Something V2 — https://developer.qualcomm.com/software/ai-datasets/something-something, then decode by index (we use `decord`; a self-maintained
sequential counter drifts silently if the decoder ever skips a frame).

### Known limitations

- Every episode here has an instruction. These episodes come from VITRA-1M's official
  segmentation, which already drops the ones a captioner marks `N/A`; we measured 0
  instruction-less episodes in this release.
- Paraphrase count averages 2.09, not a fixed number. Past 3 the model starts inventing;
  a sentence with no prepositional phrase honestly supports only one or two.
- Verified: every episode's stored frame count matches its index rows, and no index entry
  points at a missing episode.

---

<!-- COLLECTION_TABLE_START -->
## The collection

Every dataset we have taken through this pipeline, with what is published today.
All repos live under [`MIT-Media-Lab`](https://huggingface.co/MIT-Media-Lab) and are
**annotations only — no images, no video**.

| dataset | episodes | training samples | our contribution | size | HF |
|---|---:|---:|---|---:|---|
| **EPIC-KITCHENS-100** | 149,570 | 4,019,534 | episodes + text | 8.70 GB | [`epic30-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/epic30-annotations-v1) |
| **EgoTouch** | 111,159 | 3,687,389 | episodes + text + **tactile** | 20.12 GB | [`egotouch-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/egotouch-annotations-v1) |
| **GigaHands** | 70,486 | 2,266,087 | episodes + text | 2.89 GB | [`gigahands-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/gigahands-annotations-v1) |
| **Ego-Exo4D** | 67,051 | 1,757,474 | text only | 4.09 GB | [`egoexo4d-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/egoexo4d-annotations-v1) |
| **Something-Something V2** | 52,706 | 1,124,722 | text only | 4.63 GB | [`ssv2-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/ssv2-annotations-v1) |
| **OakInk2** | 28,264 | 1,371,721 | episodes + text | 1.92 GB | [`oakink2-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/oakink2-annotations-v1) |
| **TACO** | 23,757 | 736,136 | episodes + text | 1.34 GB | [`taco-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/taco-annotations-v1) |
| **HOT3D** | 18,590 | 780,678 | episodes + text | 1.72 GB | [`hot3d-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/hot3d-annotations-v1) |
| **Tachin** | 2,422 | 97,638 | episodes + text + **tactile** | 4.13 GB | [`tachin-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/tachin-annotations-v1) |
| **ARCTIC** | 12,610 | 425,796 | episodes + text | 0.85 GB | [`arctic-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/arctic-annotations-v1) |
| **H2O** | 5,696 | 196,941 | episodes + text | 0.40 GB | [`h2o-annotations-v1`](https://huggingface.co/datasets/MIT-Media-Lab/h2o-annotations-v1) |
| **total** | **542,311** | **16,464,116** | | **50.7 GB** | |

**episodes** = entries in `episode_frame_index.npz`, i.e. what a training run actually sees.
**training samples** = rows of `index_frame_pair`; the row number *is* the sample id.

*episodes + text* means we re-cut the source ourselves at wrist-speed minima and then wrote the
instructions. *text only* means the episodes are VITRA-1M's official segmentation, used unchanged,
and only the instructions are ours.

Episodes whose round-1 caption came back `N/A` (no object interaction) are **not** published —
they are excluded from both the archive and the index, so every episode here has a usable
instruction. That is why the published counts are below the totals we cut:

| | episodes on disk | published | dropped as `N/A` |
|---|---:|---:|---:|
| EPIC-KITCHENS-100 | 151,502 | 149,570 | 1,932 (1.3%) |
| EgoTouch | 147,386 | 111,159 | 36,227 (24.6%) |
| GigaHands | 92,365 | 70,486 | 21,879 (23.7%) |
| Ego-Exo4D | 67,051 | 67,051 | 0 |
| Something-Something V2 | 52,706 | 52,706 | 0 |
| OakInk2 | 37,692 | 28,264 | 9,427 (25.0%) |
| TACO | 26,454 | 23,757 | 2,697 (10.2%) |
| HOT3D | 22,578 | 18,590 | 3,988 (17.7%) |
| Tachin | 3,019 | 2,422 | 597 (19.8%) |
| ARCTIC | 14,821 | 12,610 | 2,211 (14.9%) |
| H2O | 7,792 | 5,696 | 2,096 (26.9%) |

`ssv2` and `egoexo4d` are 0 because VITRA-1M already dropped `N/A` upstream — their episodes
are the official segmentation, so there was nothing left for us to drop. Their on-disk counts are
slightly below VITRA-1M's published index (52,718 and 67,053) because round 2 marked a handful of
sentences `unusable` and we deleted those episodes: **12** from `ssv2`, **2** from `egoexo4d`.


### DexYCB was removed

**DexYCB was removed from this collection on 2026-08-30.** It is captured by 8 fixed RealSense
cameras around a table; its own `camera.role` field reads `allocentric` on all 15,878 episodes.
Unlike OakInk2, which ships an egocentric view alongside three allocentric ones, DexYCB has no
head-mounted camera at all, so there was nothing to filter down to.

### Not published yet

| dataset | episodes cut | where it stands |
|---|---:|---|
| HOI4D | — | source converted by a colleague; not re-cut |
<!-- COLLECTION_TABLE_END -->