Landmine detection with a mobile application

The object of the research is the detection of explosive objects in an image, with a particular focus on the identification of anti-personnel landmines. The objective of this research is to develop effective tools for the recognition of landmines. A mobile application for the recognition of explosive objects, trained on a deep learning model using landmine replicas, has been developed. The application was tested on images of actual landmines. The model utilized in the application exhibited a recall rate of 89% (calculated as the ratio of correctly identified landmines to the total number of landmines present in the image). The results indicated that the recall rate for a specific category of landmines was less than that observed for the others. The average time required for offline image recognition was 2.1 seconds. This paper presents the results of the evaluation of the effectiveness of the mobile application for landmine detection and classification. Furthermore, it describes the ways in which the application allows for the improvement of the model through the collection of data from users. It also describes the architecture and interface of the application, as well as an analysis of its potential applications in landmine recognition. The efficacy of the mobile application can be attributed to its intuitive interface, the high accuracy of the deep learning model, and the capacity to obtain user feedback promptly. The program enables not only the identification of hazardous objects but also the transmission of data for the enhancement of the model. The mobile application has the potential to be utilized for a multitude of tasks pertaining to the detection of explosive objects, in addition to enhancing the precision of the model. Furthermore, the app can be utilized in training centers for deminers and in mine-contaminated areas. The mobile application can be employed to identify unknown explosive objects and enhance the efficacy of deep learning models. The resulting models can be leveraged in the future to automate the demining process