This article presents a differential perovskite hemispherical photodetector designed for intelligent imaging and location tracking. The photodetector, with a high external quantum efficiency (up to 1000%) and low noise (10^-13 A Hz^-0.5), enables stable and large variations in signal response. By analyzing the differential light response of only 8 pixels using a computer algorithm, the device can achieve colorful imaging and computational spectral resolution of 4.7 nm in a low-cost, lensless setup. Machine learning techniques, particularly neural network fitting (NNF), are used to process the differential current signals under different applied biases, allowing for dynamic tracking of object trajectories in 3D space or 2D planes with color classification capabilities. The study demonstrates the potential of this photodetector in advanced applications such as computational spectrometers, wide-angle imaging, and intelligent detection systems. The fabrication process involves spray-coating perovskite films on hemispherical substrates, which are enhanced with amphiphilic molecules to improve device performance. The results show that the differential photodetector can effectively capture and process complex light signals, making it a promising tool for future intelligent imaging and sensing applications.This article presents a differential perovskite hemispherical photodetector designed for intelligent imaging and location tracking. The photodetector, with a high external quantum efficiency (up to 1000%) and low noise (10^-13 A Hz^-0.5), enables stable and large variations in signal response. By analyzing the differential light response of only 8 pixels using a computer algorithm, the device can achieve colorful imaging and computational spectral resolution of 4.7 nm in a low-cost, lensless setup. Machine learning techniques, particularly neural network fitting (NNF), are used to process the differential current signals under different applied biases, allowing for dynamic tracking of object trajectories in 3D space or 2D planes with color classification capabilities. The study demonstrates the potential of this photodetector in advanced applications such as computational spectrometers, wide-angle imaging, and intelligent detection systems. The fabrication process involves spray-coating perovskite films on hemispherical substrates, which are enhanced with amphiphilic molecules to improve device performance. The results show that the differential photodetector can effectively capture and process complex light signals, making it a promising tool for future intelligent imaging and sensing applications.