Incorporation of deep learning has transformed object detection in industrial environments, especially in the identification of mining spare parts, where accuracy and speed are critical to operational efficiency. This Systematic Literature Review (SLR) aims to analyze the impact of deep learning-based object detection models applied to industrial contexts. The PICO methodology was used to structure the research question, define key terms, and construct the advanced search equation. The search was carried out in the Scopus database, retrieving a total of 1,355 articles, of which 67 met the established inclusion criteria. The results show that models based on convolutional neural networks (CNNs), YOLO, and Faster R-CNN are the most widely used in spare parts detection in industrial environments, achieving significant improvements in automation, reducing human errors, and real-time traceability. It also highlights the importance of considering data quality, the availability of suitable hardware, and integration with enterprise systems. This review highlights the strategic value of deep learning in critical industrial applications.