Using two million photographs, we apply machine learning techniques to group images based on the similarity of their product and service representations. Our image classifications (IIC) perform well with industries whose value is driven by expectation, growth, and intangibility. IIC generates the highest Sharpe and Calmar ratios for the optimized portfolios that minimize risk and maximize the Sharpe ratio among the other eighteen industry classifications based on SICS, GICS, NAICS, and text similarity. The industry momentum strategies based on images yield the highest Calmar ratio. IIC mirrors the human brain’s clustering and memorizing photos better than text and numbers. The agreement within (across) an image-classified industry is highest (lowest), generating the highest impact of aggregated demand and supply on the stock price. This explains why IIC performs well in diversification and industry momentum strategies.