Figure 1. This is a caption a) Schematic of a double layer sensor pixel where each layer parallelly and independently handles a different visual task. Here, the object detection layer (ODL) detects the original object whereas the occlusion handling layer (OHL) learns and reconstructs the blocked object. As an example, the parts of the letter ‘E” is being masked by a red rectangle. b) Schematic of our sensor’s occlusion handling algorithm. Yellow frame in the Frame 1 image indicates detected object by CNN, in this case a white square. Algorithm steps numbered 1 and 2 are detailed in the text. Black frame in image of step 2 highlights the object’s reconstructed parts hidden behind a blocking red square foreground. c) Image, pixel numbering and device architecture of our proof-of-concept OHL consisting of 20 pixels. FTO stands for fluorine doped tin oxide. The yellow spheres represent TiO2 nanoparticles sensitized with photoactive dyes. d) Charge transfer mechanisms occurring in the OHL pixel during illumination.