Artifact: A Large-Scale Dataset With Artificial And Factual Images For Generalizable And Robust Synthetic Image Detection.
Md Awsafur RahmanBishmoy PaulNajibul Haque SarkerZaber Ibn Abdul HakimShaikh Anowarul FattahPublished in: ICIP (2023)
Keyphrases
- input image
- million images
- image dataset
- image data
- image collections
- image noise
- image features
- test images
- image retrieval
- partial occlusion
- image set
- original images
- pixel values
- geometric distortions
- image pixels
- web images
- image analysis
- image regions
- region of interest
- image classification
- images with ground truth
- sample images
- texture and shape features
- image matching
- edge detection
- spot detection
- object detectors
- segmentation method
- circular hough transform
- image content
- image database
- image search
- cluttered background
- gray value
- bounding box
- illumination conditions
- street view
- fundamental problem in computer vision
- image transformations
- geometric transformations
- accurate localization
- intensity variations
- orientation estimation
- annotated images
- aerial images
- single image
- scale invariant features
- photo collections
- imaging process
- segmentation algorithm
- image structure
- image annotation
- lighting conditions
- secret image
- keypoints
- gray level
- pavement distress
- complex background
- pixel level
- feature points
- image registration
- set of real images
- feature descriptors
- accurate registration
- registration errors
- contrast enhancement
- visual features
- scale space
- image quality
- road signs
- image sequences
- image representation
- color images
- high resolution
- multiscale