The presence of chemical pollutants in water poses a serious threat to aquatic ecosystems and human health, but traditional chemical analysis methods have limitations such as being time-consuming, costly, and having a limited detection range. Therefore, this paper introduces a method based on ResNet convolutional neural network to achieve rapid and low-cost detection of chemical pollutants in water and assess their potential risks to the ecosystem. The paper first obtains image information of water samples through image acquisition technology, uses the ResNet convolutional neural network model to extract features from the image, and extracts key features related to chemical pollutants from the image. Based on feature extraction, the classifier and regression model are further trained. Finally, the risks that the chemical pollutants may cause to the ecosystem are evaluated based on the types and concentrations of the identified chemical pollutants. Compared with traditional chemical analysis methods, the ResNet-based method significantly shortens the detection time, with the shortest detection time being 8 s, while the shortest for the chemical analysis method is 20.1 s. In terms of cost, the ResNet method reduces dependence on expensive laboratory equipment and professionals, reducing long-term manpower and material costs. This paper provides a rapid and low-cost method for detecting chemical pollutants in water, which overcomes the time-consuming and costly limitations of traditional chemical analysis methods and provides strong technical support for environmental protection and human health.