We propose a Lagrangian method for simultaneous, volumetric temperature and velocity measurements. As tracer particles for both quantities, we employ encapsulated thermochromic liquid crystals (TLCs). We discuss the challenges arising from color imaging of small particles and present measurements in an equilateral hexagonal-shaped convection cell of height h = 60 mm and distance between the parallel side walls w = 104 mm, which corresponds to an aspect ratio
Γ
=
1.73
. As fluid, we use a water-glycerol mixture to match the density of the TLC particles. We propose a densely-connected neural network, trained on calibration data, to predict the temperature for individual particles based on their particle image and position in the color camera images, which achieves uncertainties below 0.2 K over a temperature range of 3 K. We use Shake-the-Box to determine the 3D position and velocity of the particles and couple it with our temperature measurement approach. We validate our approach by adjusting a stable temperature stratification and comparing our measured temperatures with the theoretical results. Finally, we apply our approach to thermal convection at Rayleigh number Ra =
3.4
×
10
7
and Prandtl number Pr = 10.6. We can visualize detaching plumes in individual temperature and convective heat transfer snapshots. Furthermore, we demonstrate that our approach allows us to compute statistics of the convective heat transfer and briefly validate our results against the literature.