README.md

---
dataset_info:
  features:
  - name: uid
    dtype: string
  - name: video_id
    dtype: string
  - name: start_second
    dtype: float32
  - name: end_second
    dtype: float32
  - name: caption
    dtype: string
  - name: fx
    dtype: float32
  - name: fy
    dtype: float32
  - name: cx
    dtype: float32
  - name: cy
    dtype: float32
  - name: vid_w
    dtype: int32
  - name: vid_h
    dtype: int32
  - name: annotation
    list:
    - name: mano_params
      struct:
      - name: global_orient
        list: float32
      - name: hand_pose
        list: float32
      - name: betas
        list: float32
    - name: is_right
      dtype: bool
    - name: keypoints_3d
      list: float32
    - name: keypoints_2d
      list: float32
    - name: vertices
      list: float32
    - name: box_center
      list: float32
    - name: box_size
      dtype: float32
    - name: camera_t
      list: float32
    - name: focal_length
      list: float32
  splits:
  - name: train
    num_examples: 241912
  - name: test
    num_examples: 5108
configs:
- config_name: default
  data_files:
  - split: train
    path: EgoHaFL_train.csv
  - split: test
    path: EgoHaFL_test.csv
license: mit
language:
- en
pretty_name: EgoHaFL:Egocentric 3D Hand Forecasting Dataset with Language Instruction
size_categories:
- 200K<n<300K
tags:
- embodied-ai
- robotic
- egocentric
- 3d-hand
- forecasting
- hand-pose
---

# **EgoHaFL: Egocentric 3D Hand Forecasting Dataset with Language Instruction**

**EgoHaFL** is a dataset designed for **egocentric (first-person) 3D hand forecasting** with accompanying **natural language instructions**.
It contains short video clips, text descriptions, camera intrinsics, and detailed MANO-based 3D hand annotations.
The dataset supports research in **3D hand forecasting**, **hand pose estimation**, **hand–object interaction understanding**, and **video–language modeling**.

![Demo GIF](EgoHaFL.gif)

[Paper on Arxiv](https://arxiv.org/pdf/2511.18127)

---

## 📦 **Dataset Contents**

### **1. Metadata CSV Files**

* `EgoHaFL_train.csv`
* `EgoHaFL_test.csv`

Each row corresponds to one sample and contains:

| Field            | Description                                |
| ---------------- | ------------------------------------------ |
| `uid`            | Unique sample identifier                   |
| `video_id`       | Source video identifier                    |
| `start_second`   | Start time of the clip (seconds)           |
| `end_second`     | End time of the clip (seconds)             |
| `caption`        | Natural language instruction / description |
| `fx`, `fy`       | Camera focal lengths                       |
| `cx`, `cy`       | Principal point                            |
| `vid_w`, `vid_h` | Original video resolution                  |

---

### **2. 3D Hand Annotations (EgoHaFL_lmdb)**

The folder `EgoHaFL_lmdb` stores all 3D annotations in **LMDB format**.

* **Key**: `uid`
* **Value**: a **list of length 16**, representing uniformly sampled frames across a **3-second video segment**.

Each of the 16 elements is a dictionary containing:

* `mano_params`

  * `global_orient (n, 1, 3 ,3)`
  * `hand_pose (n, 15, 3, 3)`
  * `betas (n, 10)`
* `is_right (n,)`
* `keypoints_3d (n, 21, 3)`
* `keypoints_2d (n, 21, 2)`
* `vertices (n, 778, 3)`
* `box_center (n, 2)`
* `box_size (n,)`
* `camera_t (n, 3)`
* `focal_length (n, 2)`

Here, `n` denotes the number of hands present in each frame, which may vary across frames. When no hands are detected, the dictionary is empty.

---

## 🌳 **Annotation Structure (Tree View)**

Below is the hierarchical structure for a single annotation entry (`uid → 16-frame list → per-frame dict`):

```
<uid>
└── list (length = 16)
    ├── [0]
    │   ├── mano_params
    │   │   ├── global_orient
    │   │   ├── hand_pose
    │   │   └── betas
    │   ├── is_right
    │   ├── keypoints_3d
    │   ├── keypoints_2d
    │   ├── vertices
    │   ├── box_center
    │   ├── box_size
    │   ├── camera_t
    │   └── focal_length
    ├── [1]
    │   └── ...
    ├── [2]
    │   └── ...
    └── ...
```

---

## 🎥 **Source of Video Data**

The video clips used in **EgoHaFL** originate from the **Ego4D V1** dataset.
For our experiments, we use the **original-length videos compressed to 224p resolution** to ensure efficient storage and training.

Official Ego4D website:
🔗 **[https://ego4d-data.org/](https://ego4d-data.org/)**

---

## 🧩 **Example of Use**

For details on how to load and use the EgoHaFL dataset,
please refer to the **dataloader implementation** in our GitHub repository:

🔗 **[https://github.com/ut-vision/SFHand](https://github.com/ut-vision/SFHand)**

---

## 🧠 **Supported Research Tasks**

* Egocentric 3D hand forecasting
* Hand motion prediction and trajectory modeling
* 3D hand pose estimation
* Hand–object interaction understanding
* Video–language multimodal modeling
* Temporal reasoning with 3D human hands