README.md

Thermo Presence Dataset

The Thermo Presence Dataset consists of thermal images with resolution 24x32 px and corresponding annotations in the form of X and Y coordinates. Recorded data are in hdfs directory. Raw data are available in the cloud shared folder.

Collected frames are divided into sequences which then are split into training, validation, and test sets as follows:

training_dirs: [
    "006__11_44_59", "007__11_48_59", "008__11_52_59", 
    "009__11_57_00", "000__14_15_19", "001__14_19_19", 
    "002__14_23_19", "003__14_27_20", "004__14_31_20", 
    "012__15_03_21", "013__15_07_21", "014__15_11_21", 
    "015__15_15_21", "016__15_19_21", "011__13_38_20", 
    "012__13_42_20", "013__13_46_21", "007__13_22_20"
]

validation_dirs: [
    "004__13_10_20", "014__13_50_21", "005__14_35_20", 
    "006__14_39_20", "007__14_43_20", "008__14_47_20"
]

test_dirs: [
    "008__13_26_20", "009__14_51_20", "010__14_55_20", 
    "011__14_59_20", "015__13_54_21"
]

The table below shows the summary of training, validation, and test datasets considering the number of people in the frame.

	0	1	2	3	4	5	Total
Training	99	105	2984	3217	1953	114	8472
Validation	0	139	631	1691	225	139	2825
Test	162	83	211	341	1235	315	2347

Python examples

Load dataset file

import pandas as pd


df = pd.read_hdf('./dataset/hdfs/007__13_22_20.h5')
df.head()

Visualize an example thermal image with corresponding annotations

import pandas as pd
import matplotlib.pyplot as plt


df = pd.read_hdf('./dataset/hdfs/007__13_22_20.h5')

idx = 100

x = np.array(df.iloc[idx]['points'])[:, 0]
y = np.array(df.iloc[idx]['points'])[:, 1]
frame = df.iloc[idx]['data']

plt.imshow(frame)
plt.scatter(x=x, y=y, s=200, c='red', marker='x')
plt.show()