JRXIE/WoodThermalDatasets

Download the Dataset

The best way to download the dataset is using the huggingface_hub library.

from huggingface_hub import snapshot_download

local_dir = "./DATA/"
snapshot_download(
    repo_id="JRXIE/WoodThermalDatasets", 
    repo_type="dataset",
    local_dir=local_dir
)

Visualize the Dataset

Once the dataset is downloaded, you can load the images, raw temperature matrices, and raw bed matrices. The following script applies Z-score normalization to the thermal data and visualizes them.

import cv2
import numpy as np
from numpy import genfromtxt
import matplotlib.pyplot as plt

base = 100
DATA_DIR = './DATA/Poplar'
images_A = []
images_B = []
thermals_A = []
thermals_B = []
beds_A = []
beds_B = []
N = 12

print("[info] Loading files from disk...")
for i in range(0, N):
    # Load Images (Convert BGR to RGB for matplotlib)
    filepath_A = f'{DATA_DIR}/Image/PD_{i+100}_A.jpg'
    image_A = cv2.imread(filepath_A, cv2.IMREAD_COLOR) 
    images_A.append(cv2.cvtColor(image_A, cv2.COLOR_BGR2RGB))
    
    filepath_B = f'{DATA_DIR}/Image/PD_{i+100}_B.jpg'
    image_B = cv2.imread(filepath_B, cv2.IMREAD_COLOR)
    images_B.append(cv2.cvtColor(image_B, cv2.COLOR_BGR2RGB))

    # Load Thermals
    filepath_A = f'{DATA_DIR}/Temperature/PD_{i+100}_A.csv'
    thermal_A = genfromtxt(f'{filepath_A}', delimiter=',')
    filepath_B = f'{DATA_DIR}/Temperature/PD_{i+100}_B.csv'
    thermal_B = genfromtxt(f'{filepath_B}', delimiter=',')
    thermals_A.append(thermal_A)
    thermals_B.append(thermal_B)

    # Load Bed
    filepath_A = f'{DATA_DIR}/Bed/PD_{i+100}_A.csv'
    bed_A = genfromtxt(f'{filepath_A}', delimiter=',')
    filepath_B = f'{DATA_DIR}/Bed/PD_{i+100}_B.csv'
    bed_B = genfromtxt(f'{filepath_B}', delimiter=',')
    beds_A.append(bed_A)
    beds_B.append(bed_B)

# Convert to numpy arrays
thermals_A = np.array(thermals_A)
thermals_B = np.array(thermals_B)
beds_A = np.array(beds_A)
beds_B = np.array(beds_B)

print("[info] Normalizing data...")
thermals_A = (thermals_A - np.mean(thermals_A, axis=(1, 2), keepdims=True)) / np.std(thermals_A, axis=(1, 2), keepdims=True)
thermals_B = (thermals_B - np.mean(thermals_B, axis=(1, 2), keepdims=True)) / np.std(thermals_B, axis=(1, 2), keepdims=True)

beds_A = (beds_A - np.mean(beds_A, axis=(1, 2), keepdims=True)) / np.std(beds_A, axis=(1, 2), keepdims=True)
beds_B = (beds_B - np.mean(beds_B, axis=(1, 2), keepdims=True)) / np.std(beds_B, axis=(1, 2), keepdims=True)

print("[info] Generating plot...")
plt.figure(figsize=(12, 3), constrained_layout=True)

for i in range(12):
    plt.subplot(3, 12, i+1)
    plt.imshow(images_A[i], interpolation='none')
    plt.axis('off')

    plt.subplot(3, 12, i+1+12)
    plt.imshow(thermals_A[i], cmap='inferno', vmin=-3.0, vmax=3.0, interpolation='none')
    plt.axis('off')

    plt.subplot(3, 12, i+1+24)
    plt.imshow(beds_A[i], cmap='inferno', vmin=-3.0, vmax=3.0, interpolation='none')
    plt.axis('off')

plt.savefig('Poplar.png', dpi=300)
print("[info] Visualization saved as Poplar.png!")
plt.show()
plt.close()

Citation

If you use this dataset in your research, please cite the following paper.

@article{xie2026physics, 
  title={Physics-Informed Modeling for Wood Thermal Analysis and Prediction}, 
  author={Xie, Jingren and Buckthal, Alex John and O'Connor, Ryan Anthony and Foged, Isak Worre and Papadopoulos, Dim P}, 
  journal={arXiv preprint arXiv:2606.23402}, 
  year={2026} 
}