Consider using the FLIR Thermal Dataset

[maxbbraun]

The FLIR ADAS Thermal Dataset has >10k thermal images, some of which include people. It comes with annotations, but those are bounding boxes around the whole body, not just the face.

It could be useful to get the subset of images with people annotated (e.g. like this) and mix it in with the other training data.

[maxbbraun]

The images also come in 16-bit versions, allowing for the possibility of applying the same pre-processing used at inference time and getting a closer match between dataset and inference inputs.

[maxbbraun]

• Created the dataset: flir-adas-faces
• Integrated the dataset: 674cc7ae408a776da1139bf24cbb1f1ed086ed14, 437b2eee7e95b01ab31c1a3aca2cfc2bdb3c5198, 8157bd8176c66f429265a1ead5b93a6886e2a159, 7fc2efceace695ba9fe6257877019e23b7eeeb2e, 7da56ca01271ca77d4f54c8b0dbb17532f4e23e6
• Trained new models: fbd9dce6bb0cf3d5362ae5861722215a0a7de5f8

[maxbbraun]

Another potentially useful dataset: FLIR European Regional Thermal Dataset