Hugging Face's logo

Datasets:
Skywork
/
unipic_nano_2images

Tasks:
Image-to-Image
Text-to-Image
Modalities:
Text
Formats:
json
Languages:
English
Size:
10K - 100K
ArXiv:
Tags:
image-composition
multi-image
image-fusion
image-editing
unipic
2-image-input
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
input_images
listlengths
2
2
instruction
stringlengths
49
407
output_image
stringlengths
19
23
[ "30/0.png", "30/1.png" ]
A man from Image1 is standing on a surfboard from Image2, riding the ocean waves under a bright blue sky.
30/fusion_result.png
[ "22/0.png", "22/1.png" ]
A woman in a red dress from Image1 is confidently watching the television displaying the "ELEMENT" logo from Image2, creating a focused and engaged atmosphere.
22/fusion_result.png
[ "15/0.png", "15/1.png" ]
A woman with long red hair and glasses from Image1 is holding a jar of beans from Image2, smiling warmly, creating a cheerful and inviting scene.
15/fusion_result.png
[ "24/0.png", "24/1.png" ]
A woman in a red dress from Image1 is elegantly holding an apple from Image2, creating a vibrant and fresh scene.
24/fusion_result.png
[ "1/0.png", "1/1.png" ]
A woman in a pink top from Image1 is holding a purple mop from Image2, creating a cheerful and active scene.
1/fusion_result.png
[ "32/0.png", "32/1.png" ]
A woman in a blue dress from Image1 is standing under a glowing street lamp from Image2, creating a romantic and dreamy atmosphere.
32/fusion_result.png
[ "38/0.png", "38/1.png" ]
A woman from Image1 is sitting on the transparent chair from Image2, creating a modern and stylish scene.
38/fusion_result.png
[ "26/0.png", "26/1.png" ]
A woman in a cream coat from Image1 is elegantly playing the blue violin from Image2, creating a sophisticated and artistic ensemble.
26/fusion_result.png
[ "36/0.png", "36/1.png" ]
A person with words written on their face from Image1 is using a tissue from Image2, creating a thoughtful and introspective scene.
36/fusion_result.png
[ "12/0.png", "12/1.png" ]
A woman with short blonde hair from Image1 is elegantly wearing the colorful Egyptian-themed backpack from Image2, creating a stylish and adventurous ensemble.
12/fusion_result.png
[ "16/0.png", "16/1.png" ]
A woman in a pink suit from Image1 is elegantly sitting on the wooden bench from Image2, creating a stylish and serene scene.
16/fusion_result.png
[ "34/0.png", "34/1.png" ]
A man in a blue jacket from Image1 is confidently playing the golden saxophone from Image2, creating a dynamic and stylish scene.
34/fusion_result.png
[ "2/0.png", "2/1.png" ]
A person with bold makeup and a spiked collar from Image1 is confidently wearing the black helmet from Image2, creating a striking and edgy look.
2/fusion_result.png
[ "46/0.png", "46/1.png" ]
A man from Image1 is wearing a yellow towel hat from Image2, creating a playful and relaxed look.
46/fusion_result.png
[ "44/0.png", "44/1.png" ]
A woman in a blue and white dress from Image1 is confidently riding the skateboard from Image2, creating a dynamic and stylish scene.
44/fusion_result.png
[ "48/0.png", "48/1.png" ]
A woman from Image1 is comfortably sitting on the checkered sofa from Image2, holding a book from the bookshelf, creating a cozy and relaxed atmosphere.
48/fusion_result.png
[ "45/0.png", "45/1.png" ]
A man from Image1 is standing confidently in front of a modern metallic building from Image2, creating a professional and urban atmosphere.
45/fusion_result.png
[ "41/0.png", "41/1.png" ]
A man and woman lying on a patterned carpet from Image1 are elegantly holding a bright yellow handbag from Image2, creating a stylish and sophisticated scene.
41/fusion_result.png
[ "52/0.png", "52/1.png" ]
A woman from Image1 is joyfully riding a bike while holding a slice of pepperoni pizza from Image2, creating a playful and delicious scene.
52/fusion_result.png
[ "51/0.png", "51/1.png" ]
A woman from Image1 is standing on a road with "WAY ROD" painted on it from Image2, wearing a red lace dress and a gold cross necklace, with trees lining the road in the background.
51/fusion_result.png
[ "54/0.png", "54/1.png" ]
A woman from Image1 is elegantly wearing the pink ballet shoes from Image2, creating a graceful and poised appearance.
54/fusion_result.png
[ "11/0.png", "11/1.png" ]
A woman from Image1 is elegantly wearing the pink hat from Image2, creating a stylish and sophisticated ensemble.
11/fusion_result.png
[ "29/0.png", "29/1.png" ]
A woman from Image1 is elegantly wearing a green facial mask from Image2, creating a serene and self-care focused look.
29/fusion_result.png
[ "59/0.png", "59/1.png" ]
A man from Image1 is casually drinking from the wine glass in Image2, creating a relaxed and sophisticated atmosphere.
59/fusion_result.png
[ "17/0.png", "17/1.png" ]
A woman from Image1 is elegantly holding the small golden purse from Image2, surrounded by fluffy cats, creating a cozy and sophisticated scene.
17/fusion_result.png
[ "56/0.png", "56/1.png" ]
A man in a light blue suit from Image1 is confidently playing the silver flute from Image2, creating a harmonious and elegant scene.
56/fusion_result.png
[ "60/0.png", "60/1.png" ]
A person with tattoos from Image1 is comfortably lying on a couch, holding a steaming glass of red beverage heated by a small electric burner from Image2, creating a cozy and relaxed atmosphere.
60/fusion_result.png
[ "39/0.png", "39/1.png" ]
A person dressed as Santa Claus from Image1 is stylishly wearing the rugged watch from Image2, adding a touch of sophistication to the festive outfit.
39/fusion_result.png
[ "64/0.png", "64/1.png" ]
A young person with a big smile from Image1 is joyfully playing with a curious brown dog from Image2, creating a heartwarming and lively scene.
64/fusion_result.png
[ "49/0.png", "49/1.png" ]
A man from Image1 is using the camera while wearing the blue textured napkin from Image2 as a unique headband, creating a quirky and creative look.
49/fusion_result.png
[ "67/0.png", "67/1.png" ]
A woman in a white lace dress from Image1 is elegantly reading an open book from Image2, creating a serene and thoughtful atmosphere.
67/fusion_result.png
[ "27/0.png", "27/1.png" ]
A man with glasses and a beard from Image1 is casually playing golf, holding the golf club from Image2, with a focused expression, creating a relaxed and sporty scene.
27/fusion_result.png
[ "5/0.png", "5/1.png" ]
A man in a white shirt and jeans from Image1 is confidently holding a colorful basketball and wearing a professional video camera from Image2 on his shoulder, capturing his game highlights.
5/fusion_result.png
[ "69/0.png", "69/1.png" ]
A man and a woman from Image1 are enjoying a night out, with the man wearing a stylish black sweater and glasses, and the woman in a vibrant red dress. The man is holding a saxophone from Image2, playing a lively tune as they share a joyful moment together, creating a harmonious and elegant scene.
69/fusion_result.png
[ "72/0.png", "72/1.png" ]
A woman from Image1 is standing and presenting, while sitting on a wooden chair from Image2, creating a dynamic and engaging scene.
72/fusion_result.png
[ "76/0.png", "76/1.png" ]
A woman with red lipstick from Image1 is elegantly holding a book from Image2, with the word "EMPIRE" prominently displayed in the background.
76/fusion_result.png
[ "80/0.png", "80/1.png" ]
A woman from Image1 is elegantly sitting on the leopard print bench from Image2, creating a stylish and sophisticated scene.
80/fusion_result.png
[ "83/0.png", "83/1.png" ]
A woman in a gold sequin dress from Image1 is elegantly holding a red apple from Image2, creating a stylish and sophisticated ensemble.
83/fusion_result.png
[ "88/0.png", "88/1.png" ]
A woman from Image1 is elegantly wearing the silver high-heeled shoes from Image2, creating a stylish and sophisticated ensemble.
88/fusion_result.png
[ "25/0.png", "25/1.png" ]
A woman with long red hair from Image1 is standing by the water, holding a black Audix microphone from Image2 close to her mouth, as if about to sing or speak, creating a serene and artistic scene.
25/fusion_result.png
[ "89/0.png", "89/1.png" ]
A woman from Image1 is gracefully riding the horse from Image2, creating a dynamic and harmonious scene.
89/fusion_result.png
[ "79/0.png", "79/1.png" ]
A woman from Image1 is elegantly wearing the black baseball helmet from Image2 and holding the wooden baseball bat, creating a stylish and sporty ensemble.
79/fusion_result.png
[ "86/0.png", "86/1.png" ]
A woman from Image1 is enjoying a slice of pizza from Image2, savoring the cheesy, ham, and basil flavors.
86/fusion_result.png
[ "92/0.png", "92/1.png" ]
A woman in a black strapless top and leather pants from Image1 is elegantly wearing the black bow tie from Image2, creating a stylish and sophisticated ensemble.
92/fusion_result.png
[ "47/0.png", "47/1.png" ]
A woman in a blue sports jersey from Image1 is confidently sitting on the red chair from Image2, creating a dynamic and stylish scene.
47/fusion_result.png
[ "43/0.png", "43/1.png" ]
A woman in a flowing floral dress from Image1 is gracefully painting with the turquoise paintbrush from Image2, creating an artistic and elegant scene.
43/fusion_result.png
[ "93/0.png", "93/1.png" ]
A woman in a red shirt from Image1 is holding a baby in a pink dress and gently sipping from a white ribbed cup from Image2, creating a warm and tender moment.
93/fusion_result.png
[ "21/0.png", "21/1.png" ]
A woman in a black shirt and hat from Image1 is stylishly playing the saxophone from Image2, creating a dynamic and engaging scene.
21/fusion_result.png
[ "65/0.png", "65/1.png" ]
A man in a blue vest from Image1 is elegantly wearing the gold bracelet from Image2, creating a stylish and sophisticated ensemble.
65/fusion_result.png
[ "81/0.png", "81/1.png" ]
A man in a yellow hard hat and safety vest from Image1 is holding a green can from Image2, smiling as he takes a break from work.
81/fusion_result.png
[ "62/0.png", "62/1.png" ]
A man in a navy polo shirt and gray pants from Image1 is energetically playing the saxophone from Image2, creating a lively and dynamic scene.
62/fusion_result.png
[ "95/0.png", "95/1.png" ]
A man in a suit from Image1 is holding a wooden baseball bat from Image2, standing confidently with a young girl beside him, both smiling.
95/fusion_result.png
[ "101/0.png", "101/1.png" ]
A man with a beard and a black turtleneck from Image1 is sitting at a grand piano from Image2, playing the piano with a smile.
101/fusion_result.png
[ "104/0.png", "104/1.png" ]
A woman from Image1 is elegantly wearing a stylish oven mitt from Image2, creating a unique and sophisticated ensemble.
104/fusion_result.png
[ "63/0.png", "63/1.png" ]
A woman in a red dress from Image1 is elegantly wearing the blue hiking boots from Image2, creating a stylish and adventurous ensemble.
63/fusion_result.png
[ "105/0.png", "105/1.png" ]
A man in a patterned shirt from Image1 is sitting on a couch and brushing the globe from Image2 with the brush labeled "Friendly Products," creating a unique and engaging scene.
105/fusion_result.png
[ "112/0.png", "112/1.png" ]
A woman from Image1 is stretching on a colorful mat in a park, wearing a blue patterned shirt and black leggings, while holding a glass of red wine from Image2 in one hand, creating a relaxed and serene atmosphere.
112/fusion_result.png
[ "74/0.png", "74/1.png" ]
A woman in a white shirt and blue jeans from Image1 is elegantly wearing the crown from Image2, creating a stylish and sophisticated ensemble.
74/fusion_result.png
[ "18/0.png", "18/1.png" ]
A man in a workshop setting is playing a grand piano, featuring the main subject from Image1 interacting with the piano from Image2.
18/fusion_result.png
[ "118/0.png", "118/1.png" ]
A woman in a teal blouse from Image1 is elegantly holding a beige River Island handbag from Image2, creating a stylish and sophisticated ensemble.
118/fusion_result.png
[ "116/0.png", "116/1.png" ]
The elderly man from Image1 is watching the television from Image2, creating a scene of quiet contemplation.
116/fusion_result.png
[ "115/0.png", "115/1.png" ]
A woman from Image1 is elegantly holding a silver microphone with a USB connector from Image2, creating a stylish and sophisticated ensemble.
115/fusion_result.png
[ "96/0.png", "96/1.png" ]
A woman in a red shirt from Image1 is confidently wearing the black high-heeled sandals from Image2, creating a stylish and bold ensemble.
96/fusion_result.png
[ "119/0.png", "119/1.png" ]
A man and woman from Image1 are standing together, with the woman elegantly wearing a dress adorned with the teal paint from Image2, creating a stylish and artistic ensemble.
119/fusion_result.png
[ "10/0.png", "10/1.png" ]
A woman in a stylish outfit from Image1 is elegantly holding the gel cleanser bottle from Image2, creating a chic and sophisticated look.
10/fusion_result.png
[ "125/0.png", "125/1.png" ]
A man from Image1 is confidently holding the blue golf club from Image2, ready to take a swing on the course.
125/fusion_result.png
[ "94/0.png", "94/1.png" ]
A woman in a striped sweater from Image1 is elegantly holding a wooden-handled screwdriver from Image2, creating a stylish and sophisticated ensemble.
94/fusion_result.png
[ "91/0.png", "91/1.png" ]
A woman from Image1 is elegantly holding a slice of pizza from Image2, creating a casual and delightful dining scene.
91/fusion_result.png
[ "126/0.png", "126/1.png" ]
A woman in a white dress from Image1 is elegantly holding a pomegranate from Image2, with seeds scattered around her, creating a striking and sophisticated scene.
126/fusion_result.png
[ "77/0.png", "77/1.png" ]
A woman in a white saree from Image1 is elegantly singing into the microphone from Image2, creating a harmonious and artistic scene.
77/fusion_result.png
[ "102/0.png", "102/1.png" ]
A woman from Image1 is elegantly holding the golden fan from Image2, creating a stylish and sophisticated ensemble.
102/fusion_result.png
[ "114/0.png", "114/1.png" ]
A woman from Image1 is elegantly wearing the sparkling choker necklace from Image2, creating a stylish and sophisticated ensemble.
114/fusion_result.png
[ "135/0.png", "135/1.png" ]
A woman in a green blazer from Image1 is elegantly sipping wine from the glass in Image2, creating a sophisticated and relaxed atmosphere.
135/fusion_result.png
[ "136/0.png", "136/1.png" ]
A woman in a beige trench coat from Image1 is elegantly wearing the green high-heeled shoes from Image2, creating a stylish and sophisticated ensemble.
136/fusion_result.png
[ "113/0.png", "113/1.png" ]
A woman from Image1 is elegantly wearing the teal leather shoes from Image2, creating a stylish and sophisticated ensemble.
113/fusion_result.png
[ "143/0.png", "143/1.png" ]
A man from Image1 is casually eating a banana from Image2, taking a bite with a relaxed expression, creating a simple and everyday moment.
143/fusion_result.png
[ "139/0.png", "139/1.png" ]
A woman from Image1 is elegantly standing beside a towering stone structure from Image2, with a massive cloud formation rising dramatically behind her, creating a striking and surreal scene.
139/fusion_result.png
[ "140/0.png", "140/1.png" ]
A man from Image1 is elegantly wearing the gold ring with a large teal gemstone from Image2, creating a stylish and sophisticated ensemble.
140/fusion_result.png
[ "100/0.png", "100/1.png" ]
A woman in a vibrant pink gown from Image1 is elegantly wearing the black high-heeled shoes from Image2, creating a stylish and sophisticated ensemble.
100/fusion_result.png
[ "142/0.png", "142/1.png" ]
A man with curly hair and glasses from Image1 is sitting on a floral-patterned sofa, holding a blue tube of "Blue Blue" face wash from Image2, with a serene expression, creating a relaxed and casual atmosphere.
142/fusion_result.png
[ "40/0.png", "40/1.png" ]
A woman from Image1 is elegantly holding a wine glass from Image2, sipping red wine, creating a sophisticated and relaxed atmosphere.
40/fusion_result.png
[ "103/0.png", "103/1.png" ]
A woman from Image1 is gently holding a child while both are playfully interacting with a violin from Image2, creating a joyful and musical moment.
103/fusion_result.png
[ "107/0.png", "107/1.png" ]
A young man with curly hair from Image1 is enthusiastically singing into the vintage microphone from Image2, creating a lively and engaging performance.
107/fusion_result.png
[ "145/0.png", "145/1.png" ]
A woman from Image1 is sweeping the ground with the purple broom from Image2, creating a magical and serene atmosphere.
145/fusion_result.png
[ "147/0.png", "147/1.png" ]
A woman in a blue coat from Image1 is elegantly wearing the black visor from Image2, creating a stylish and sophisticated ensemble.
147/fusion_result.png
[ "4/0.png", "4/1.png" ]
A woman in a bright orange sweater from Image1 is elegantly sitting on the luxurious cream sofa from Image2, creating a stylish and sophisticated scene.
4/fusion_result.png
[ "149/0.png", "149/1.png" ]
A woman from Image1 is standing under a street lamp from Image2, holding a sign that reads "MAST STREET," creating a charming and whimsical scene.
149/fusion_result.png
[ "148/0.png", "148/1.png" ]
A man with tattoos from Image1 is elegantly sweeping the floor with the broom from Image2, creating a unique and intriguing scene.
148/fusion_result.png
[ "137/0.png", "137/1.png" ]
A woman in a blue dress from Image1 is elegantly wearing the tan blazer from Image2, creating a stylish and sophisticated ensemble.
137/fusion_result.png
[ "84/0.png", "84/1.png" ]
A woman in a brown coat from Image1 is confidently riding a bright yellow surfboard from Image2, creating a stylish and adventurous scene.
84/fusion_result.png
[ "133/0.png", "133/1.png" ]
A woman from Image1 is joyfully holding a child while playing the saxophone from Image2, creating a lively and heartwarming scene.
133/fusion_result.png
[ "70/0.png", "70/1.png" ]
A man with gray hair and a beard from Image1 is confidently wearing the beige shirt and large green ring from Image2, creating a stylish and sophisticated ensemble.
70/fusion_result.png
[ "154/0.png", "154/1.png" ]
A woman in a blue jacket from Image1 is elegantly wearing the jade bracelet from Image2, creating a harmonious and natural look.
154/fusion_result.png
[ "156/0.png", "156/1.png" ]
A man from Image1 is casually holding a tennis racket from Image2, with tennis balls scattered around, creating a relaxed and sporty scene.
156/fusion_result.png
[ "164/0.png", "164/1.png" ]
A woman in a pink dress from Image1 is elegantly playing the trumpet from Image2, creating a stylish and sophisticated ensemble.
164/fusion_result.png
[ "132/0.png", "132/1.png" ]
A woman from Image1 is gracefully playing the acoustic guitar from Image2, creating a harmonious and engaging scene.
132/fusion_result.png
[ "166/0.png", "166/1.png" ]
A woman from Image1 is elegantly wearing the gold wristwatch from Image2, creating a stylish and sophisticated ensemble.
166/fusion_result.png
[ "168/0.png", "168/1.png" ]
A young girl from Image1 is playfully holding carrots above her head, while wearing a whimsical outfit made from the colorful label of the HIRMEL can from Image2, creating a fun and imaginative scene.
168/fusion_result.png
[ "138/0.png", "138/1.png" ]
A woman in a yellow top from Image1 is holding the flashlight from Image2, creating a focused and practical scene.
138/fusion_result.png
[ "109/0.png", "109/1.png" ]
A man with curly hair and a beard from Image1 is elegantly wearing the dark suit and red tie from Image2 while playing the golden trumpet, creating a stylish and sophisticated ensemble.
109/fusion_result.png
End of preview. Expand in Data Studio

Skywork/unipic_nano_2images: A Multi-Image Composition Dataset

⚡ Quick Start

The image archive is split into multiple parts for easier downloading. To reconstruct and extract:

# Step 1: Concatenate split files into a single zip
cat nano-banana.part_* > nano-banana-2images.zip

# Step 2: Extract the images
unzip nano-banana-2images.zip

📖 Overview

UniPic-Nano-2Images is a high-quality multi-image composition dataset containing 41,812 samples designed for training image fusion and composition models. Each sample consists of 2 input images and 1 output image, where elements from both input images are seamlessly combined based on natural language instructions. This dataset is part of the UniPic series and has been used in UniPic3 for training advanced multi-image composition models.

🎨 Demo: Multi-Image Composition

This dataset enables models to intelligently combine subjects and objects from two separate images into a single coherent output based on natural language instructions:

UniPic3 Multi-Image Composition Demo

Example: Person from Image1 combined with objects/scenes from Image2 to create a seamless composition

🎯 Key Features

  • Multi-Image Input: Each sample uses exactly 2 input images for composition
  • Diverse Composition Tasks: Covers 18 different composition scenarios including wearing accessories, holding objects, playing instruments, and more
  • High Quality: 41,812 carefully curated samples with natural language instructions
  • Production Ready: Used in UniPic3 for real-world multi-image composition applications
  • Simple Format: Clean JSON format with straightforward input/output structure

📊 Dataset Statistics

Task Type Count Percentage Description
Wearing Jewelry/Accessories 9,721 23.2% Person wearing watches, bracelets, necklaces, rings, earrings
Eating Food 7,704 18.4% Person eating various food items
Holding Other Object 6,519 15.6% Person holding miscellaneous objects
Playing Instrument 4,671 11.2% Person playing musical instruments (guitar, violin, saxophone, etc.)
Sitting on Furniture 4,323 10.3% Person sitting on chairs, sofas, benches
Holding Drink/Container 2,175 5.2% Person holding glasses, cups, bottles
Wearing Headwear 1,653 4.0% Person wearing hats, caps, helmets
Holding Everyday Object 1,094 2.6% Person holding books, phones, cameras, umbrellas
Holding Food 921 2.2% Person holding food items
Wearing Eyewear 643 1.5% Person wearing glasses, sunglasses
Holding Flowers 553 1.3% Person holding flowers, bouquets
Holding Musical Instrument 546 1.3% Person holding (not playing) instruments
Holding Bag 349 0.8% Person holding bags, purses, backpacks
Riding Vehicle 297 0.7% Person riding bikes, surfboards, skateboards
Standing in Scene 233 0.6% Person standing in various backgrounds/scenes
Other Interaction 188 0.4% Other composition types
Using Tool/Device 181 0.4% Person using various tools or devices
Drinking Beverage 41 0.1% Person drinking beverages
Total 41,812 100% All multi-image composition samples

📁 Dataset Structure

Data Format

Each sample in the dataset is a JSON object with the following structure:

{
  "input_images": ["path/to/image1.png", "path/to/image2.png"],
  "instruction": "A woman from Image1 is elegantly wearing the gold bracelet from Image2, creating a stylish ensemble.",
  "output_image": "path/to/fusion_result.png"
}

Field Descriptions

  • input_images: List of exactly 2 input image paths

    • Image1: Typically contains the main subject (person)
    • Image2: Typically contains the object/scene to be composed

  • instruction: Natural language description of how to combine the two images, following the pattern:

    • Subject description from Image1
    • Action/interaction verb (wearing, holding, sitting on, playing, etc.)
    • Object description from Image2
    • Scene/atmosphere description

  • output_image: Path to the composed output image

Composition Pattern

The dataset follows a consistent composition pattern:

[Subject from Image1] + [Action] + [Object from Image2] → [Fused Output]

Example instructions:

  • "A man from Image1 is standing on a surfboard from Image2, riding the ocean waves under a bright blue sky."
  • "A woman in a cream coat from Image1 is elegantly playing the blue violin from Image2, creating a sophisticated and artistic ensemble."
  • "A woman from Image1 is comfortably sitting on the checkered sofa from Image2, holding a book, creating a cozy atmosphere."

🚀 Usage

Loading the Dataset

Using Hugging Face Datasets 

from datasets import load_dataset

# Load the dataset from Hugging Face
dataset = load_dataset("Skywork/unipic_nano_2images", split="train")

# Access a sample
sample = dataset[0]
print(f"Input images: {sample['input_images']}")
print(f"Instruction: {sample['instruction']}")
print(f"Output image: {sample['output_image']}")

Direct JSON Loading 

import json

# Load from local JSONL file
samples = []
with open("unipic_nano_2images.jsonl", "r", encoding="utf-8") as f:
    for line in f:
        sample = json.loads(line.strip())
        samples.append(sample)

print(f"Total samples: {len(samples)}")  # 41,812

Using PyTorch DataLoader 

from torch.utils.data import Dataset, DataLoader
from PIL import Image
import json

class UniPicNano2ImagesDataset(Dataset):
    def __init__(self, jsonl_path, image_root):
        self.samples = []
        with open(jsonl_path, "r", encoding="utf-8") as f:
            for line in f:
                self.samples.append(json.loads(line.strip()))
        self.image_root = image_root
    
    def __len__(self):
        return len(self.samples)
    
    def __getitem__(self, idx):
        sample = self.samples[idx]
        # Load input images
        img1 = Image.open(f"{self.image_root}/{sample['input_images'][0]}")
        img2 = Image.open(f"{self.image_root}/{sample['input_images'][1]}")
        # Load output image
        output = Image.open(f"{self.image_root}/{sample['output_image']}")
        
        return {
            "input_images": [img1, img2],
            "instruction": sample["instruction"],
            "output_image": output
        }

dataset = UniPicNano2ImagesDataset("unipic_nano_2images.jsonl", "images/")
dataloader = DataLoader(dataset, batch_size=8, shuffle=True)

Filtering by Task Type 

import json
import re

def categorize_sample(instruction):
    """Categorize a sample based on its instruction."""
    instruction = instruction.lower()
    
    if 'wearing' in instruction:
        if any(x in instruction for x in ['watch', 'bracelet', 'necklace', 'ring', 'earring']):
            return 'Wearing Jewelry/Accessories'
        elif any(x in instruction for x in ['hat', 'cap', 'helmet']):
            return 'Wearing Headwear'
        elif any(x in instruction for x in ['glasses', 'sunglasses']):
            return 'Wearing Eyewear'
    elif 'holding' in instruction:
        return 'Holding Object'
    elif 'playing' in instruction:
        return 'Playing Instrument'
    elif 'sitting' in instruction:
        return 'Sitting on Furniture'
    elif 'eating' in instruction:
        return 'Eating Food'
    
    return 'Other'

# Filter samples by category
with open("unipic_nano_2images.jsonl", "r") as f:
    samples = [json.loads(line) for line in f]

jewelry_samples = [s for s in samples if categorize_sample(s['instruction']) == 'Wearing Jewelry/Accessories']
print(f"Jewelry samples: {len(jewelry_samples)}")  # ~9,721

🔬 Task Categories

1. Wearing Compositions (28.7%)

Person from Image1 wearing items from Image2:

  • Jewelry/Accessories: Watches, bracelets, necklaces, rings, earrings
  • Headwear: Hats, caps, helmets, visors
  • Eyewear: Glasses, sunglasses

2. Holding Compositions (23.8%)

Person from Image1 holding objects from Image2:

  • Everyday Objects: Books, phones, cameras, umbrellas
  • Food Items: Fruits, snacks, meals
  • Flowers: Bouquets, roses
  • Containers: Glasses, cups, bottles
  • Bags: Purses, backpacks, handbags

3. Activity Compositions (40.6%)

Person from Image1 performing activities with items from Image2:

  • Playing Instruments: Guitar, violin, saxophone, trumpet, piano
  • Sitting on Furniture: Chairs, sofas, benches
  • Eating Food: Various food items
  • Riding Vehicles: Bikes, surfboards, skateboards

4. Scene Compositions (6.9%)

Person from Image1 placed in scenes/backgrounds from Image2:

  • Standing in Scene: Various backgrounds and environments
  • Using Tools: Various tools and devices

🎓 Applications

This dataset is designed for training and evaluating:

  • Multi-Image Composition Models: Learn to seamlessly combine elements from multiple images
  • Instruction-Following Vision Models: Models that follow natural language composition instructions
  • Image Editing Models: Fine-grained control over image composition
  • Subject-Object Fusion: Intelligent blending of subjects with objects/scenes

🔗 Related Work

This dataset is part of the UniPic dataset series:

📝 Citation

If you use this dataset in your research, please cite:

@misc{wei2026skyworkunipic30unified,
  title={Skywork UniPic 3.0: Unified Multi-Image Composition via Sequence Modeling}, 
  author={Hongyang Wei and Hongbo Liu and Zidong Wang and Yi Peng and Baixin Xu and Size Wu and Xuying Zhang and Xianglong He and Zexiang Liu and Peiyu Wang and Xuchen Song and Yangguang Li and Yang Liu and Yahui Zhou},
  year={2026},
  eprint={2601.15664},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2601.15664}, 
}

@misc{wang2025skyworkunipicunifiedautoregressive,
  title={Skywork UniPic: Unified Autoregressive Modeling for Visual Understanding and Generation}, 
  author={Peiyu Wang and Yi Peng and Yimeng Gan and Liang Hu and Tianyidan Xie and Xiaokun Wang and Yichen Wei and Chuanxin Tang and Bo Zhu and Changshi Li and Hongyang Wei and Eric Li and Xuchen Song and Yang Liu and Yahui Zhou},
  year={2025},
  eprint={2508.03320},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2508.03320}, 
}

📄 License

Please refer to the license terms on the Hugging Face dataset page.

Downloads last month
151
Size of the auto-converted Parquet files:
2.77 MB
Number of rows:
41,812

Collection including Skywork/unipic_nano_2images

Papers for Skywork/unipic_nano_2images