Sign-language and video templates
Three gallery groups are relevant. Every template is editable on the Labeling Setup step — treat the defaults below as starting points.
Sign Language Collection
All three use a single <VideoRecord> pointed at the prompt, capped at
30 seconds, front camera, tracking off. Change maxDuration, add
skeleton="hands" (or face, all), or swap the prompt object as needed.
| Template | Prompt | Task data | Config |
|---|---|---|---|
| Sign this prompt (video corpus) | Written word / phrase / sentence | text | <Text name="prompt" value="$text"/> + <VideoRecord name="recording" toName="prompt" maxDuration="30"/> |
| Sign a picture (video corpus) | Image of the concept | image | <Image name="prompt" value="$image" maxWidth="600"/> + the same recorder |
| Prompted question (spontaneous signing) | Written question, answered in the signer's own signs | text | <Text name="prompt" value="$text"/> + the same recorder |
Intended use, from the template notes: isolated-sign and continuous-signing corpora and lexical elicitation (Sign this prompt); picture-based elicitation independent of written-language literacy, iconicity and variation studies (Sign a picture); continuous / discourse data, non-manual-marker elicitation, natural variation (Prompted question — allow a longer clip for a full answer). Framing advice in all three: head to waist, even front lighting, plain background, hands inside the frame.
Behaviour of the recorder itself: Recording.
Sign Language Annotation
| Template | Tags | Task data | What it does |
|---|---|---|---|
| Utterances and translation (first pass) | <VideoTiers tiers="utterance,translation:utterance" timelock="on" mode="span"/> | video | The first pass over a new recording, and the right start for annotators new to corpus work: mark each utterance, translate it, no glossing. Produces the parent segments the hand and non-manual tiers later hang inside. .eaf export. |
| Gloss and translate (sign video) | <Video frameRate="25"> + <TimelineLabels> (Sign / Fingerspelling / Non-manual) + per-region <Choices> hand (dominant / non-dominant / both symmetric / both asymmetric), per-region <TextArea> gloss, per-region <TextArea> translation, per-region <Choices> handshape (19-form starter inventory + Other) | video | Single-timeline segmentation with an ID-gloss, spoken-language translation, which hand articulates, and dominant handshape per segment. Region-based rather than tiered, so the hand is a per-region field instead of its own tier. Ordinary timeline regions — no .eaf. |
| Multi-tier annotation (sign video) | <VideoTiers tiers="utterance,rh:utterance,lh:utterance,mouthing:rh,brow:utterance,head:utterance,translation:utterance" timelock="on" pose="$pose" mode="span"/> | video, optional pose | Stacked, time-locked tiers with frame-stepping; pose overlay when the task carries pre-computed keypoints (remove pose otherwise); .eaf export. Add lexicon="<kb id>" to gloss the hand tiers against a shared ID-gloss lexicon. |
Why the tiers are shaped this way — and what an ID-gloss is — is on How annotation works.
Details and the .eaf round-trip: Annotation.
Videos (general purpose)
Not sign-specific, but the same <Video> family — useful for QC passes or
auxiliary labels on a sign corpus.
| Template | Tags | Task data | What it does |
|---|---|---|---|
| Video Classification | <Video> + <Choices> (Blurry / Sharp) | video | One choice per clip — e.g. a quick usable / unusable pass. |
| Video Frame Classification | <TimelineLabels> (Movement / Still / Slow Motion) + <Video timelineHeight="120"> | video | Label spans of frames on the timeline. |
| Video Object Tracking | <Labels> (Man / Woman / Other) + <Video framerate="25.0"> + <VideoRectangle> | video | Tracked bounding boxes with keyframes; the frame rate applies to every video in the project. |
| Video Timeline Segmentation | <Video sync="audio"> + <Audio sync="video"> + <Labels> on the audio | video_url | Segments marked on a waveform kept in sync with the (muted) video — the "audio sync trick". |
Picking between the groups
- Collecting new footage → Sign Language Collection.
- Glossing footage you already have → Sign Language Annotation (multi-tier when manual and non-manual channels must align; single timeline for a first pass).
- Everything else on video (classification, tracking, chaptering) → Videos.
The gallery entry for every group is in Templates.