Datasets:

Voxel51
/

Egocentric_10K_Evaluation

Error code:   FeaturesError
Exception:    ArrowInvalid
Message:      Schema at index 1 was different: 
cls: string
custom_metrics: null
ytrue: list<item: string>
ypred: list<item: string>
confs: list<item: null>
weights: null
ytrue_ids: list<item: string>
ypred_ids: list<item: string>
classes: list<item: string>
missing: string
vs
cls: string
custom_metrics: null
ytrue: list<item: double>
ypred: list<item: double>
confs: list<item: null>
ids: list<item: string>
missing: null
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 249, in compute_first_rows_from_streaming_response
                  iterable_dataset = iterable_dataset._resolve_features()
                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 3496, in _resolve_features
                  features = _infer_features_from_batch(self.with_format(None)._head())
                                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2257, in _head
                  return next(iter(self.iter(batch_size=n)))
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2461, in iter
                  for key, example in iterator:
                                      ^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 1952, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 1974, in _iter_arrow
                  yield from self.ex_iterable._iter_arrow()
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 563, in _iter_arrow
                  yield new_key, pa.Table.from_batches(chunks_buffer)
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "pyarrow/table.pxi", line 5039, in pyarrow.lib.Table.from_batches
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
              pyarrow.lib.ArrowInvalid: Schema at index 1 was different: 
              cls: string
              custom_metrics: null
              ytrue: list<item: string>
              ypred: list<item: string>
              confs: list<item: null>
              weights: null
              ytrue_ids: list<item: string>
              ypred_ids: list<item: string>
              classes: list<item: string>
              missing: string
              vs
              cls: string
              custom_metrics: null
              ytrue: list<item: double>
              ypred: list<item: double>
              confs: list<item: null>
              ids: list<item: string>
              missing: null

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Dataset Card for Egocentric_10K_Evaluation

This is a FiftyOne dataset with 30000 samples.

Installation

If you haven't already, install FiftyOne:

pip install -U fiftyone

Usage

import fiftyone as fo
from fiftyone.utils.huggingface import load_from_hub

# Load the dataset
# Note: other available arguments include 'max_samples', etc
dataset = load_from_hub("Voxel51/Egocentric_10K_Evaluation")

# Launch the App
session = fo.launch_app(dataset)

Dataset Details

Dataset Description

Egocentric-10K-Evaluation is a benchmark evaluation set and analysis protocol for large-scale egocentric (first-person) video datasets, focused on measuring hand visibility and active manipulation in real-world, in-the-wild scenarios, especially relevant for robotics, computer vision, and AI agent training on manipulation tasks.[1][2][3]

Curated by: builddotai
Shared by : builddotai
License: Apache 2.0

Dataset Sources

Repository: https://huggingface.co/datasets/builddotai/Egocentric-10K-Evaluation

Uses

Direct Use

This dataset is intended for benchmarking egocentric video data with respect to hand presence and active object manipulation, enabling standardized analysis, dataset comparison, and the development/evaluation of perception and robotics models centered on real-world human skill tasks.

Dataset Structure

Egocentric-10K-Evaluation consists of 10,000 sampled frames from factory egocentric video and comparable samples from other major datasets (Ego4D, EPIC-KITCHENS); each sample includes JSON metadata, hand label annotations (count 0, 1, or 2), and a binary label for presence/absence of active manipulation. The splits are standardized; additional metadata includes dataset, worker, and video index references.

Dataset Creation

Curation Rationale

To create a standardized benchmark for hand visibility and manipulation, facilitating research on manipulation-heavy tasks in robotics and AI using real industrial and skill-focused footage.

Source Data

Data Collection and Processing

The evaluation set comprises frames drawn from the primary Egocentric-10K dataset (real-world factory footage collected via head-mounted cameras), as well as standardized samples from open egocentric datasets Ego4D and EPIC-KITCHENS for comparison. Data is provided in 1080p, 30 FPS H.265 MP4 format, with structured JSON metadata and hand/manipulation annotations.

Who are the source data producers?

Egocentric-10K’s original video data was produced by real factory workers wearing head-mounted cameras, performing natural work-line activities. Annotation was performed following strict guidelines as described in the evaluation schema.

Annotations

Annotation process

Each sampled frame is annotated for number of visible hands (0/1/2; detailed rules provided) and whether the hands are engaged in active manipulation (“yes”/“no” per explicit definition). The annotation schema and rules are detailed in the benchmark documentation.

Citation

Downloads last month: 1,634