Computer vision algorithms are becoming more widely used in the agricultural domain. Edge detection algorithms are used to detect plant or seed contours. These contours identify regions of interest and delimit the size of plants or seeds from a single perspective. By combining contours from different angles, other plant phenotypes, such as volume, can be computed. This paper delves into the potential of edge detection, specifically contour detection, in the domain of seed phenotyping. Contour detection is significantly influenced by image quality, with color saturation due to ambient light and background color playing important roles. This paper expands traditional contour detection algorithms to address their limitations and improve their accuracy. Images are captured using different physical characteristics and image backgrounds. A consistent background color allows for more contrast with the seeds being analyzed. This paper compares various traditional detection methods with convolutional neural network (CNN) detection methods. The analysis aims to highlight the strengths and weaknesses of each technique, offering valuable insights for seed researchers because contour detection is essential in supporting non-destructive measurement of seed volume. This volume can be used to compute the density of irregularly-shaped seeds.