Color histogram equalization and fine-tuning to improve expression recognition of (partially occluded) faces on sign language datasets
Fabrizio Nunnari, Alakshendra Singh, Patrick Gebhard · 2025
Figure 1: Example pictures of our procedure to classify Ekman emotions.The original picture is first normalized in size, rotation, and color profile.Then, we perform classification on the full face, only top part, and only bottom part.On the left: examples from the AffectNet dataset.On the right: examples from the PHOENIX sign language dataset.