Fig. 9
From: High-throughput soybean seeds phenotyping with convolutional neural networks and transfer learning
An example of the visual results using unsupervised segmentation methods on our real-world soybean seed images. a real-world soybean seeds test image captured by Kinect v2 outdoor, b real-world soybean seeds test image captured by iPhone 6Â s plus at indoor scene in the daytime with fluorescent light, c real-world soybean seeds test image captured by iPhone 6Â s plus at indoor scene at night with fluorescent light. The first row was the input soybean seeds test image, the second row was the results of the thresholding and contour extraction method, and the third row was the results of watershed algorithm