video video 60 60 | setting stringlengths 4 19 | activity stringlengths 18 56 | argus_summary stringlengths 127 190 | viewpoint stringclasses 1
value | camera stringclasses 1
value | audio bool 1
class | duration_s float64 60 60 | width int64 1.28k 1.92k | height int64 720 1.08k | fps float64 30 30 | num_segments int64 27 85 | num_tasks int64 1 7 | data_issues int64 0 0 | annotated bool 1
class | tasks listlengths 1 7 | segments listlengths 27 85 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Backyard | sweeping a concrete path with a push broom | The person uses a long-handled push broom to sweep loose dirt and grit along a concrete garden path, changing working angle and grip as they move. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 74 | 1 | 0 | true | [
{
"start_s": 0,
"end_s": 59.97,
"task": "Sweep loose dirt and grit from the concrete garden path",
"objects": [
"push broom",
"loose dirt and grit",
"concrete garden path"
],
"outcome": "unclear",
"success_predicate": "The intended stretch of path is cleared of loose su... | [
{
"start_s": 0,
"end_s": 0.5,
"arm": "both",
"action": "sweep broom across concrete",
"object": "push broom",
"destination": "dusty path",
"spatial_relation": "bristles against concrete",
"contribution": "advancing",
"progress": 0.05,
"hands_visible": true,
"hands_wearing... | |
Bicycle | riding a bicycle with a bag at the handlebar | The person rides and steers a bicycle, briefly takes hands off the grips, and moves an off-white tote from beside the left thigh to a held position against the left handlebar. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 32 | 2 | 0 | true | [
{
"start_s": 0,
"end_s": 59.97,
"task": "Ride and steer the bicycle along the paved lane",
"objects": [
"bicycle",
"handlebar"
],
"outcome": "unclear",
"success_predicate": "The ride reaches its stopping point or destination.",
"completed_at_s": null
},
{
"start_s... | [
{
"start_s": 0,
"end_s": 1.5,
"arm": "both",
"action": "hold grips and steer while riding",
"object": "handlebar",
"destination": "along paved lane",
"spatial_relation": null,
"contribution": "advancing",
"progress": 0.05,
"hands_visible": true,
"hands_wearing": "bare",
... | |
Construction site | applying sealing foam with a foam gun | The person repeatedly shakes and reorients a Soudafoam can attached to a dispensing gun, alternating between two-handed support and right-handed shaking while moving beside unfinished walls. | first_person | head-mounted phone, first person | false | 60 | 1,280 | 720 | 30 | 29 | 1 | 0 | true | [
{
"start_s": 0,
"end_s": 59.97,
"task": "Prepare a foam can on its dispensing gun through repeated shaking and reorientation",
"objects": [
"Soudafoam expanding-foam can",
"foam dispensing gun"
],
"outcome": "unclear",
"success_predicate": "The agitation/preparation is finish... | [
{
"start_s": 0,
"end_s": 3,
"arm": "both",
"action": "carry and reorient the foam assembly while stepping along the wall",
"object": "foam can and dispensing gun",
"destination": "farther along the wall",
"spatial_relation": "held in front of the body",
"contribution": "advancing",
... | |
Craft studio | weaving on a small hand loom | The person hand-weaves a narrow burgundy band, selecting and tensioning warp threads, passing a yarn shuttle, working the rows with wooden tools, and checking the band with a ruler. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 85 | 7 | 0 | true | [
{
"start_s": 0,
"end_s": 11.5,
"task": "Finish a weft pass and work the new row into place",
"objects": [
"warp threads",
"woven burgundy band",
"yarn-wound weaving shuttle",
"fork-ended wooden weaving stick",
"wooden rigid heddle"
],
"outcome": "success",
"... | [
{
"start_s": 0,
"end_s": 0.5,
"arm": "right",
"action": "draw shuttle outward",
"object": "yarn-wound weaving shuttle",
"destination": "right side of warp",
"spatial_relation": "clear of the warp",
"contribution": "advancing",
"progress": 0.1,
"hands_visible": true,
"hand... | |
Farm | carrying fresh grass and feeding goats | The person gathers three armfuls of fresh green fodder, carries them into a goat enclosure, and distributes them on the floor, briefly offering the final armful directly to the goats. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 52 | 3 | 0 | true | [
{
"start_s": 0,
"end_s": 10.5,
"task": "Carry and distribute the first armful of green fodder inside the goat enclosure",
"objects": [
"source pile of cut green fodder",
"first armful of green fodder",
"metal gate"
],
"outcome": "success",
"success_predicate": "The carr... | [
{
"start_s": 0,
"end_s": 0.5,
"arm": "both",
"action": "secure grip around an armful",
"object": "first armful of green fodder",
"destination": null,
"spatial_relation": null,
"contribution": "advancing",
"progress": 0.1,
"hands_visible": true,
"hands_wearing": "bare",
... | |
Garden | mixing in a water barrel and watering plants | The person stirs dark water in a barrel with a wooden stick, fills a watering can by immersion, and waters three garden shrubs. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 31 | 5 | 0 | true | [
{
"start_s": 0,
"end_s": 22.5,
"task": "Stir the barrel water and put the stirring stick aside",
"objects": [
"open brown barrel",
"dark water",
"wooden stirring stick"
],
"outcome": "success",
"success_predicate": "The water has been visibly agitated and the stick is r... | [
{
"start_s": 0,
"end_s": 18.5,
"arm": "both",
"action": "stir repeatedly with the stick",
"object": "wooden stirring stick",
"destination": "barrel water",
"spatial_relation": "moving through the water",
"contribution": "advancing",
"progress": 0.85,
"hands_visible": true,
... | |
Home | sorting and folding clothes at a wardrobe | The person closes an underwear drawer, puts black, tan and brown garments onto wardrobe shelves, then prepares a blue garment for storage in another drawer. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 70 | 5 | 0 | true | [
{
"start_s": 0,
"end_s": 2.5,
"task": "Finish arranging underwear and close its drawer",
"objects": [
"underwear",
"underwear drawer"
],
"outcome": "success",
"success_predicate": "The underwear remains inside the drawer and the drawer is closed.",
"completed_at_s": 2
}... | [
{
"start_s": 0,
"end_s": 0.5,
"arm": "both",
"action": "settle lace garments",
"object": "underwear",
"destination": "underwear drawer",
"spatial_relation": "inside the drawer",
"contribution": "advancing",
"progress": 0.7,
"hands_visible": true,
"hands_wearing": "bare",
... | |
Shop / stockroom | sorting and arranging shop stock | The person carries groceries to a shopfront counter, arranges coffee and baking ingredients, retrieves another stock tray from inside, and begins arranging boxed tea. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 75 | 4 | 0 | true | [
{
"start_s": 0,
"end_s": 8,
"task": "Carry the loaded grocery crate to the exterior display counter",
"objects": [
"black plastic lattice crate",
"packaged groceries"
],
"outcome": "success",
"success_predicate": "The loaded crate rests on the exterior counter without hand su... | [
{
"start_s": 0,
"end_s": 6.5,
"arm": "both",
"action": "carry loaded crate through doorway and along shopfront",
"object": "black plastic lattice crate",
"destination": "red exterior counter",
"spatial_relation": "held in front of body",
"contribution": "advancing",
"progress": 0... | |
Interior renovation | working with a cordless drill | The person works along floor-level baseboard trim with a cordless drill, adjusts the tool, and selects screws for successive fixing points. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 40 | 3 | 0 | true | [
{
"start_s": 0,
"end_s": 26.5,
"task": "Work on the fixing at the baseboard corner",
"objects": [
"cordless drill/driver",
"baseboard trim"
],
"outcome": "unclear",
"success_predicate": "The corner fixing or hole is completed to its intended final condition.",
"completed_... | [
{
"start_s": 0,
"end_s": 5.5,
"arm": "both",
"action": "hold and operate drill at corner",
"object": "cordless drill/driver",
"destination": "baseboard corner",
"spatial_relation": "bit against corner fixing point",
"contribution": "advancing",
"progress": 0.3,
"hands_visible... | |
Painting table | decorating long fingernails with a fine brush (nail art) | The person loads a fine brush with paint and adds detailed brushwork to a long left-hand nail, repeatedly changing the hand's angle and refreshing the brush. | first_person | head-mounted phone, first person | false | 60 | 1,920 | 1,080 | 30 | 27 | 1 | 0 | true | [
{
"start_s": 0,
"end_s": 59.97,
"task": "Add painted detail to the long nails of the left hand",
"objects": [
"left-hand nails",
"fine paintbrush",
"white paint jar",
"palette plate"
],
"outcome": "unclear",
"success_predicate": "The intended nail decoration is fi... | [
{
"start_s": 0,
"end_s": 3,
"arm": "right",
"action": "work brush tip at the open paint jar",
"object": "fine paintbrush",
"destination": "white paint jar",
"spatial_relation": "at the jar opening",
"contribution": "advancing",
"progress": 0.08,
"hands_visible": true,
"ha... |
End of preview. Expand in Data Studio
Egocentric hands-on work: 12-setting sample (Weply Labs)
12 one-minute clips (12 min) of people doing real hands-on work, filmed from a head-mounted phone camera (first person), each from a different setting. Recorded by Weply contributors; every clip comes with dense annotations of what each hand does.
License: CC BY-NC 4.0. Free to use, share and adapt for non-commercial purposes, with credit to Weply Labs; contact Weply Labs for commercial use and for the full collection.
| Clip | Setting | Activity | Tasks | Hand segments |
|---|---|---|---|---|
backyard.mp4 |
Backyard | sweeping a concrete path with a push broom | 1 | 74 |
bicycle.mp4 |
Bicycle | riding a bicycle with a bag at the handlebar | 2 | 32 |
construction_site.mp4 |
Construction site | applying sealing foam with a foam gun | 1 | 29 |
craft_studio.mp4 |
Craft studio | weaving on a small hand loom | 7 | 85 |
farm.mp4 |
Farm | carrying fresh grass and feeding goats | 3 | 52 |
garden.mp4 |
Garden | mixing in a water barrel and watering plants | 5 | 31 |
home.mp4 |
Home | sorting and folding clothes at a wardrobe | 5 | 70 |
indoor_workplace.mp4 |
Shop / stockroom | sorting and arranging shop stock | 4 | 75 |
interior_renovation.mp4 |
Interior renovation | working with a cordless drill | 3 | 40 |
painting_studio.mp4 |
Painting table | decorating long fingernails with a fine brush (nail art) | 1 | 27 |
restaurant_kitchen.mp4 |
Restaurant kitchen | making sushi rolls | 3 | 78 |
sewing.mp4 |
Textile craft | knotting a fringe by hand at a sewing table | 1 | 50 |
Video
1080p (the construction clip is 720p at source), 30 fps, H.264, 60 s each, no audio. The camera is a phone worn on the head, looking at the wearer's hands from eye level; nobody else's face appears in the clips. Files carry no capture metadata.
Annotations
Produced with Pantheon's open-source Argus, self-hosted, from the video alone (sampled at 2 frames per second, so boundaries are accurate to about half a second), and spot-checked by Weply.
annotations/segments.jsonl: per-hand action timeline. One row per segment:file_name,start_s,end_s,arm(left / right / both hands),action(free-text verb phrase),object,destination,spatial_relation,contribution(advancing / wasteful / idle),progress(0 to 1),hands_visible,hands_wearing(bare, gloves, ...).annotations/tasks.jsonl: the distinct activities in each clip withoutcomeandsuccess_predicate.annotations/<clip>.json: the full labels per clip: scene objects, timeline, tasks, key events, object state changes, scene graph, recoveries, data issues and a short performance review.videos/metadata.jsonl: one row per clip with setting, activity, Argus summary, video properties, and that clip'stasksandsegmentsembedded, so the dataset viewer shows each video next to its annotations.
from datasets import load_dataset
clips = load_dataset("weply-labs/egocentric_handwork_sample", split="train") # one row per clip: video, metadata, tasks, segments
segments = load_dataset("json", data_files="hf://datasets/weply-labs/egocentric_handwork_sample/annotations/segments.jsonl", split="train")
- Downloads last month
- 165