README.md

metadata

license: cc-by-nc-4.0
pretty_name: VoxMem EPIC-KITCHENS share
tags:
  - egocentric
  - epic-kitchens
  - depth
  - camera-pose
  - video-qa

VoxMem EPIC-KITCHENS share

This share holds derived data of the VoxMem project, built from EPIC-KITCHENS-100. The videos are (c) the EPIC-KITCHENS team under CC BY-NC 4.0, and this share keeps that license: it is for non-commercial research. The original MP4s are not here.

Current version: 2026-10-18 (v3). It holds:

• the question set (frozen 2026-10-12), both tiers (accurate and rough), including the questions built with Gemini;
• one label row per question;
• the trainable lists;
• the token cache, 599 videos (the 572 of the previous version + 27 new ones, poses from EPIC-Fields only);
• the hand masks, the reference-frame tokens and their exact-frame 3D, and the fps table.
• the frames the tokens were made from (added 2026-10-21).

The voxels are not shared. The training machine builds them from the token cache with one script.

To train, use the code, not this page:

• the guide is docs/TRAIN_ELSEWHERE_EPIC_V3.md in the VoxMem repo, branch hdepic-gen-epic-questions;
• bash scripts/kit/download_epic_v3.sh downloads, checks and unpacks everything below into place.

Every path here is the path under the VoxMem data root (src/voxmem/paths.py), except the two reftok-4b / refgeo-4b tars, which unpack into <code>/results/cache/hdepic/.

v3 (2026-10-18)

path	what
epic-kitchens/tokens/qwen3vl-4b_512px/<video>.npz	599 files (3.43 TB), one per video. Each holds the Qwen3-VL-4B visual tokens (144 per 512x288 frame), similarity features, 3D points, camera pose, depth and the frame-rule table. Every frame is chosen by the frame rule: each wanted 1 fps frame exactly, else the nearest frame with a camera pose within +-0.5 s, else that second is dropped. sha256 in epic-kitchens/tokens/tokens_MANIFEST.tsv and SHA256SUMS_tokens.txt.
epic-kitchens/questions/questions_2026-10-18.tar	the question set: 2026-10-18/<type>/{accurate,rough}.jsonl for 25 question types, plus LIST.{csv,md} (also loose as LIST_2026-10-18.{csv,md}: per type, its counts, whether Gemini was involved, its sources) and reference_images.json
epic-kitchens/questions/labels_2026-10-18.jsonl	one row per question. The main columns: tier, gemini, split, coverage, n_frames, scene_id, voxel_tokens, est_tokens, trainable, reason (the first rule failed) and trainable_ref_train.
epic-kitchens/questions/train_2026-10-18.json + _split.json	the 68,465 trainable questions (64,439 train / 4,026 val, on 27 validation videos)
epic-kitchens/questions/train_2026-10-18_reftrain.json + _split.json	the 67,070 of them whose every reference image has 3D (for --ref_pos train)
epic-kitchens/hand_masks/hand_masks_2026-10-18.tar	box + palm hand masks, 599 videos each (unpacks to epic-kitchens/hand_masks/{box,palm}/). They supersede the loose 572-video hand_masks/{box,palm}/ files, whose rows are identical.
epic-kitchens/meta/fps_table_2026-10-18.tar	Videos/<P>/{frames,durations}.txt for the 599 videos: the voxel builder's fps table
epic-kitchens/reference_frames/reftok-4b_2026-10-18.tar	24,727 reference-image token files (144 x 2560, or 192 x 2560 for the two 4:3 videos)
epic-kitchens/reference_frames/refgeo-4b_2026-10-18.tar	23,015 exact-frame 3D files, one per reference image: per-token xyz and depth, camera pose, source. They come from the frame's EPIC-Fields pose, or a pose interpolated within 0.5 s, plus DA3 depth; otherwise from the nearest posed cached frame within 0.5 s.
epic-kitchens/reference_frames/ref512_jpg_2026-10-18.tar	the 512 px reference images the tokens were made from (needed only to re-bake them)
epic-kitchens/frames/<video>.tar	599 tars (179.1 GB, 1,708,038 JPEGs, <video>/<frame:06d>.jpg; bytes, md5 and sha256 in epic-kitchens/frames/frames_MANIFEST_2026-10-18.tsv): the undistorted 992 px frames (pinhole; 992x557, 992x743 for the two 4:3 videos P12_05 / P12_06) the tokens were made from. They hold every frame of the token cache (its depth, intrinsics and camera pose are in the token file) and every exact 1 fps frame the questions ask for; epic-kitchens/frames/frames_index_2026-10-18.tsv.gz says which is which (exact_1fps, rule_frame). 542 wanted frames do not exist (one past a video's last frame, or a corrupt last frame). Not fetched by download_epic_v3.sh: only native-Qwen training or vision-encoder fine-tuning needs them. They replace the 2026-10-06 tars (572 videos, a subset), which remain in this repo's history.
epic-kitchens/v3_MANIFEST.tsv	bytes, md5 and sha256 of every v3 file above except the tokens and frames

Renamed on 2026-10-19 so every v3 file carries one date. The question files were named after the day the questions were frozen: questions_2026-10-12.tar, LIST_2026-10-12.{csv,md}, labels_2026-10-12.jsonl and train_1012[_reftrain][_split].json. Only the names changed. Every loose file is byte-identical, and the tar holds the same 65 files under the folder 2026-10-18/ instead of 2026-10-12/. The old files remain in this repo's history.

HD-EPIC reference geometry (added 2026-10-20)

path	what
hd-epic/reference_frames/hdepic_refgeo-4b_2026-10-18.tar	2,847 files (22 MB), one per reference image of the HD-EPIC VQA questions (2,650 questions). Each holds the camera pose (Aria SLAM world), the depth of each token and each token's MEAN 3D point (12 x 12 tokens of a 384 px, 95-degree rectified frame). md5 65381813477968f5deef7aebb04e88e1

• Who needs it: only someone TESTING on HD-EPIC a model trained with ref_pos=train. Unpack it into <code>/results/cache/hdepic/ (it creates refgeo-4b/).
• One recording is empty: the 51 references of P03-20240217-131219 hold NaN, because its released SLAM poses stop at 22.8 of 54.6 minutes. None of its questions is in a scored set.
• Where it comes from: HD-EPIC (Perrett et al., CVPR 2025). Cite it and follow its terms.
• The rest of the HD-EPIC test data is not public: no HD-EPIC images, videos or tokens are in this share.

Earlier versions (kept, not used by v3)

path	what
epic-kitchens/questions/questions.tar	the 2026-10-06 question set (all.json, covered.json, split_videos.json)
epic-kitchens/train/train.tar	the v1 training lists (text_only/epic_v1_*, reference_frames/epic_v1_refzs_*)
epic-kitchens/reference_frames/reference_frames.tar	the v1 reference images
epic-kitchens/raw/*.tar	EPIC-Fields poses, Intermediate-Data (per-frame poses), VISOR and the EPIC annotations
<category>/outdated-2026-09-24/	the 2026-09-24 version (old layout: its members start with task_runtime/)

MD5SUMS.txt lists every tar and loose v3 file, and SHA256SUMS_tokens.txt lists every token file. bash download.sh <DATA> checks both and unpacks the tars that belong under the data root.