Computer vision is evolving on a daily basis. Popular computer vision techniques such as image classification and object detection have been used extensively to solve a lot of computer vision problems. In image classification, an entire image is classified. Object detection extends image classification by detecting the location of individual objects present in an image. Semantic and instance segmentation is possible with five lines of code. Install PixelLib and its dependencies: Install the latest version tensorflow(tensorflow 2.0) with: pip3 install tensorflow Install opencv-python with: pip3 install opencv-python Install scikit-image with: pip3 install scikit-image Install Pillow with: pip3 install pillow Install Pixellib: pip3 install pixellib import pixellib 
 from pixellib.semantic import semantic_segmentation 

 segment_image = semantic_segmentation() 
 segment_image.load_pascalvoc_model("deeplabv3_xception_tf_dim_ordering_tf_kernels.h5") 
 segment_image.segmentAsPascalvoc("path_to_image", output_image_name = "path_to_output_image") 链接地址： https://medium.com/@olafenwaayoola/image-segmentation-with-six-lines-0f-code-acb870a462e8