| # EventActivityNet v1.0 Release Notes |
|
|
| ## Release Status |
|
|
| EventActivityNet v1.0 has completed technical validation and is suitable for public release as documented HDF5 files and scale manifests. |
|
|
| ## Highlights |
|
|
| - 3,263 validated HDF5 files. |
| - Approximately 4.36 TB total size. |
| - 5-bin event voxel representation. |
| - 200 verified action classes. |
| - Large train/validation split: 2,316 / 947. |
| - Three nested scales: Large, Medium, Small. |
| - Final production integrity audit: PASS. |
|
|
| ## Scale Summary |
|
|
| | Scale | Videos | Hours | Train | Validation | Classes | Event-friendly | |
| |---|---:|---:|---:|---:|---:|---:| |
| | Large | 3,263 | 106.94 | 2,316 | 947 | 200 | 65.31% | |
| | Medium | 1,537 | 50.00 | 1,074 | 463 | 200 | 64.80% | |
| | Small | 667 | 20.00 | 473 | 194 | 200 | 64.62% | |
|
|
| Small is a strict subset of Medium, and Medium is a strict subset of Large. |
|
|
| ## Integrity |
|
|
| The final read-only production audit confirmed: |
|
|
| - exactly 3,263 production HDF5 files; |
| - all files open successfully; |
| - zero truncated or unreadable files; |
| - zero structural warnings; |
| - every file contains `events`, `voxel_event_start`, and `voxel_event_count`; |
| - all files use `num_bins=5`; |
| - no temporary files remain in the production dataset. |
|
|
| Exact reproducibility values: |
|
|
| - total size: 4,355,745,895,245 bytes; |
| - Large duration: 106.941600381 hours; |
| - Medium duration: 50.000000128 hours; |
| - Small duration: 20.000000374 hours. |
|
|
| ## Caveats |
|
|
| - EventActivityNet v1.0 covers 200 verified action classes. Do not claim 203 classes for this release. |
| - The historical/paper Large duration of 107.3 hours is approximate relative to recovered release metadata. The verified Large duration is 106.94 hours. |
| - Medium and Small are newly generated deterministic nested v1.0 release scales, not historical original subsets. |
| - Durations in the release manifests use `src_fmt_dur` from verification metadata. |
| - Original subset generation merged train and validation before sampling. Train/validation assignments remain recoverable from ActivityNet Captions split membership. |
|
|
| ## Recommended Hugging Face Packaging |
|
|
| Preserve per-video HDF5 identity in manifests. For payload upload, tar shards around 20 GB are recommended to balance repository object count, resumability, and user access. |
|
|
| Estimated shard counts from the publication audit: |
|
|
| - 10 GB target: about 435 shards. |
| - 20 GB target: about 218 shards. |
| - 50 GB target: about 87 shards. |
|
|
| ## Final Payload Packaging |
|
|
| - 157 train tar shards. |
| - 62 validation tar shards. |
| - 219 total tar shards. |
| - 3,263 HDF5 members. |
| - Total remote tar bytes: 4,355,753,021,440. |
| - Final shard checksums are published in `metadata/shard_checksums.sha256`. |
|
|
| ## Final Annotation Metadata |
|
|
| Additional public annotation files are provided under `annotations/` and `metadata/`: |
|
|
| - `annotations/activitynet_captions.json`: ActivityNet Captions timestamped natural-language descriptions for release videos. Validation references preserve `val_1` and `val_2` separately. |
| - `annotations/activitynet_actions.json`: ActivityNet v1.3 temporal action segments and labels. |
| - `annotations/eventactivitynet_alignment.json`: EventActivityNet project-derived caption/action alignment generated using temporal IoU with midpoint-distance fallback. |
| - `annotations/annotation_issues.jsonl`: known upstream annotation quirks recorded without changing canonical values. |
| - `metadata/video_metadata.jsonl`: original-rate timing metadata, including exact rational FPS where available. |
|
|
| Captions are timestamped descriptions from ActivityNet Captions. They are not instruction-tuning examples. |
|
|
| ## Timing Clarification |
|
|
| EventActivityNet v1.0 uses original-rate, variable-FPS ActivityNet videos. The released HDF5 files were not generated from `anet_240fps_v2` or `anet_240fps_old`. |
|
|
| Implementation-derived timing: |
|
|
| - one event slice is generated for each adjacent decoded source-frame transition `(e, e + 1)`; |
| - one full `events[t]` tensor groups five adjacent-frame transitions and has shape `(5, H, W)`; |
| - for source frame count `N`, `events_T = ceil((N - 1) / 5)`; |
| - voxel `t` covers event-slice range `[5*t, 5*t + 5)`, clipped to available transitions `[0, N - 1)`; |
| - the corresponding source-frame interval is `[5*t, min(5*t + 5, N - 1)]`; |
| - with rational FPS `fps_num / fps_den`, the approximate seconds interval is `[5*t * fps_den / fps_num, min(5*t + 5, N - 1) * fps_den / fps_num]`; |
| - the final voxel may be partial, with `voxel_event_count[t]` smaller than 5; |
| - HDF5 root `fps` is the original source-frame FPS captured by OpenCV, not a 240 fps derivative; |
| - `voxel_event_start[t]` is the generated event-slice start index for voxel `t`; |
| - `voxel_event_count[t]` is the number of generated adjacent-frame slices in voxel `t`; |
| - `voxel_event_start` and `voxel_event_count` are not timestamps and not pixel-event counts. |
|
|
| For caption/action interval `[start_seconds, end_seconds]`, use original FPS to compute: |
|
|
| ```text |
| start_frame = floor(start_seconds * fps_num / fps_den) |
| end_frame = ceil(end_seconds * fps_num / fps_den) |
| t_start = max(0, floor(start_frame / 5)) |
| t_end_exclusive = min(events_T, ceil(end_frame / 5)) |
| ``` |
|
|
| Use `[t_start, t_end_exclusive)` for Python slicing, or `[t_start, t_end_exclusive - 1]` as an inclusive range when non-empty. Do not use `time_seconds = t / fps` for voxel starts; voxel start time is approximately `5 * t / fps`. |
|
|
