Context: Current smartphone models have a very interesting set of sensors such as cameras, IMUs, GPS, and environmental variables. This combination of sensors motivates the use of smartphones in scientific and service applications. One of these applications is precision agriculture, specifically drone position estimation using computer vision in GPS-denied environments for remote crop measurements. Method: This work presents the development of EVP, a vision-based position estimation system using a modern smartphone and computer vision methods. EVP consists of two software applications: an Android app (mobile station) running on a smartphone capable of controlling the drone’s flight, acquiring images, and transmitting them through a wireless network; and another application (base station) running on a Linux-based computer capable of receiving the images, processing them and running the position estimation algorithms using the acquired images. In this work, the mobile station is placed in a quadcopter. Using EVP, users can configure the mobile and base station software, execute the vision-based position estimation method, observe position graph results on the base station, and store sensor data in a database. Results: EVP was tested in three field tests: an indoor environment, an open field flight, and a field test over the Engineering Department’s square at Universidad del Valle. The root mean square errors obtained in XY were 0,166 m, 2,8 m, and 1,4 m, respectively, and they were compared against the GPS-RTK measurements. Conclusions: As a result, a vision-based position estimation system called EVP was developed and tested in realworld experiments. This system can be used in GPS-denied environments to perform tasks such as 3D mapping, pick-up and delivery of goods, object tracking, among others.