In urban logistics, analyzing urban traffic data plays an important role in achieving higher schedule reliabilityand delivery time efficiency. To increase the diversity of urban traffic data, we developed a solution for theautomated collection and analysis of different types of traffic data. These are needed to optimize and controlthe flow of traffic.The use of traditional on-road sensors (e.g., inductive loops) for collecting data is necessary but not currentlysufficient because it cannot draw any conclusions about the type of goods being transported. In this paper,we propose a framework in which different classes of delivery vehicles and types of goods being shippedare identified in road videos by deep-learning-based image recognition method. Video sequences areautomatically evaluated according to the following criteria: (i) distinguish between individual andcommercial vehicles, (ii) identify the category of commercial vehicles, for example, van, box trucks, smalltrucks, etc. (iii) identify the special features of the vehicle body (such as the name of the carrier) to classifycommercial transportation of food, general goods or package services, etc. Using this method, logisticsthroughput of a designated city or region and the peak time of goods transportation can be obtained. Thisprovides the carrier with better pre-advice and potential actions to improve transportation efficiency. For theevaluation of our framework, we collected real street videos at different time points in the main traffic arteriesof Heilbronn, Germany. In particular, the difference between traffic flow of logistics services before andduring the COVID-19 epidemic was compared. The results of implementation and testing demonstrated ahigh-precision, low-latency performance of the framework for obtaining urban logistics data.