WB (woody breast) is an economically important muscular myopathy that negatively affects the global poultry industry. In the United States, the prevalence of WB and other myopathic defects causes yearly economic losses of hundreds of millions of dollars because of product downgrading and rejection. Manual, tactile palpation combined with visual examination is the current practice for WB assessment at poultry processing facilities, which is subjective, labor-intensive, and may induce contamination. Structured-illumination reflectance imaging (SIRI) is an emerging optical imaging technique that can be potentially used for non-destructive WB assessment of poultry meat. The SIRI_PoultryWB is the first public dataset of SIRI created for WB assessment. It consists of the raw phase-shifted pattern images acquired using an in-house-assembled broadband SIRI system from poultry meat samples, as well as the resultant demodulated images [DC (direct component), AC (amplitude component), and phase difference). A total of 168 samples, including 72 normal and 96 defective with woody breast (WB), were imaged under the illumination of three phase-shifted sinusoidal patterns at eight different spatial frequencies in the range of 0.015 – 0.150 cycles/mm. The chicken fillet samples were collected from a deboning line at a commercial poultry processing plant. The WB condition of each fillet was manually assessed by trained personnel, and each fillet was assigned to one of two classes (“Normal” and “Defective”). The “Normal” samples are WB-free, while the “Defective” samples were affected by moderate or severe WB. The dataset (the unzipped SIRI_PoultryWB.7z file) contains two folders storing the raw ("RawPatternImages") and demodulated ("DemodulatedImages") image data of each class of samples. In the "RawPatternImages" folder, where there are two folders '"Defective" and "Normal", each folder each contains a set of subfolders, each of which contains 24 raw, gray-scale phase-shifted pattern images (in .tif format and of size 2048×2048 pixels and 16 bits), corresponding to the illumination of all three phase-shifted sinusoidal patterns at eight spatial frequencies. The total image number of the dataset is 4032 (that is 168 × 24), totaling the file size of about 32 GB for all the images. In the dataset, the image files are named in a consistent, descriptive fashion (Broiler_Frequency_Phase). For instance, given an image “Broiler_015Hz_PS1.tif”, it corresponds to the first phase-shifted illumination pattern at 0.015 cycles/mm, and for the image “Broiler_090Hz_PS3.tif”, it corresponds to the third phase-shifted illumination pattern at 0.090 cycles/mm. In the "DemodulatedImages" folder, there are two folders "IntensityDCAC" and "PhaseDifference". The "IntensityDCAC" contains demodulated one set of DC images and eight sets of AC images, which are categorized into different subfolders; particularly the AC images are those obtained at eight different spatial frequencies ranging from 0.015 cycles/mm to 0.150 cycles. Within each (DC or AC) subfolder, the demulated images are grouped into the "Normal" and "Defective" categories based on the WB conditions. The "PhaseDifference" folder contains the phase difference images of the two categories of samples. Details about the image demodulation are given in Cai and Lu (2025). This dataset can be used for development and evaluation of either SIRI image demodulation approaches or machine learning models for normal and WB sample classification. It was used in our recent study on sample classification using demodulated intensity and phase images with machine learning (Cai and Lu, 2025). The phase image-based model outperformed those based on intensity images, with the best overall classification accuracy of nearly 93% in distinguishing between normal and WB-affected samples. Details about the SIRI instrumentation, dataset curation, and methodologies for SIRI image processing and modeling are described in the journal article: Cai, J., Lu, Y., 2025. Assessment of woody breast in broiler breast fillets using structured-illumination reflectance imaging coupled with surface profilometry. Journal of Food Engineering, 112459. https://doi.org/10.1016/j.jfoodeng.2024.112459. If you use the dataset in published research, please consider citing the dataset or the journal article. Hopefully, you find the dataset useful.