RAISE: a raw images dataset for digital image forensics

By Duc-Tien Dang-Nguyen, Cecilia Pasquini, Valentina Conotter, Giulia Boato. 2015

592 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Law

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Law

Citations in the corpus

The Unreasonable Effectiveness of Deep Features as a Perceptual Metric (2018, methodology): “Our dataset consists of over 161k patches, derived from the MIT-Adobe 5k dataset [7] (5000 uncompressed images) for training, and the RAISE1k dataset [9] for validation.”
Real-Time Adaptive Image Compression (2017, background): “As such, we additionally present performance on the RAISE-1k dataset (Dang-Nguyen et al., 2015) which contains 1,000 raw images.”
Generative Adversarial Networks for Extreme Learned Image Compression (2018, methodology): “Data sets: We train GC models (without semantic label maps) for compression of diverse natural images using 188k images from the Open Images data set [25] and evaluate them on the widely used Kodak image compression data set [24] as well as 20 randomly selected images from the RAISE1K data set [11].”
Sparse Gradient Regularized Deep Retinex Network for Robust Low-Light Image Enhancement (2021, methodology): “We also collect 1000 raw images from RAISE [45] as normal-light images and calculate the histogram of Y channel in YCbCr.”
EnlightenGAN: Deep Light Enhancement Without Paired Supervision (2019, methodology): “We assemble a mixture of 914 low light and 1016 normal light images from several datasets released in [43, 5] and also HDR sources [15, 25], without the need to keep any pair.”
Do GANs Leave Artificial Fingerprints? (2018, methodology): “Finally, we include also two sets, from the RAISE dataset [22], of images acquired by real cameras, so as to compare the behavior of real-world and GAN fingerprints.”
Practical Full Resolution Learned Lossless Image Compression (2018, methodology): “We evaluate on 500 randomly selected images from Open Images Test , 500 randomly selected images from RAISE1k [9], as well as 100 images from the DIV2K super-resolution data set [2].”
Photo Forensics From Rounding Artifacts (2020, methodology): “The luminance channel of 1000 raw images [14] are extracted and JPEG compressed with 18 random compression settings estimated in Section 3.”
A convolutional neural network-based approach to rate control in HEVC intra coding (2017, methodology): “The images come from the UCID dataset [11] and the RAISE dataset [12].”
Fast Image Processing with Fully-Convolutional Networks (2017, methodology): “We use two image processing datasets: MIT-Adobe 5K and RAISE [12, 18].”
Reducing Complexity of HEVC: A Deep Learning Approach (2017, unknown): “To establish the CPH-Intra database, 2000 images at resolution 4928 × 3264 were selected from the Raw Images Dataset (RAISE) [35].”
MFC Datasets: Large-Scale Benchmark Datasets for Media Forensic Challenge Evaluation (2019, background): “RAISE (RAw ImageS datasEt) (2015) [44] is a collection of 8,156 high-resolution raw images using three Nikon devices.”
Noiseprint: A CNN-Based Camera Model Fingerprint (2018, methodology): “From the RAISE dataset [73], we took 100 RAW images of size 2048 × 2048, all acquired by the same device, a Nikon D7000, to ensure that they share the same digital history.”
Benchmarking Low-Light Image Enhancement and Beyond (2021, methodology): “Among them, 1000 pairs (VE-LOL-L-Syn) are synthesized from RAW images in RAISE dataset (Dang-Nguyen et al. 2015).”
Media Forensics and DeepFakes: An Overview (2020, background): “Raw images, instead, can be found in the RAISE dataset [224], composed by 8,156 images taken by 3 different cameras.”
Distinguishing computer graphics from natural images using convolution neural networks (2017, methodology): “The Photographic images are high resolution images (about 4900 × 3200 pixels) taken from the RAISE dataset [29], directly converted from RAW format to JPEG.”
Deep Retinex Decomposition for Low-Light Enhancement (2018, methodology): “1000 raw images in RAISE [4] are used to synthesize low-light images.”
GLADNet: Low-Light Enhancement Network with Global Awareness (2018, methodology): “We collect 780 raw images from RAISE [12], 700 for generating pairs for training and 80 for validation.”