【python / mxnet / gluoncv / jupyter notebook】基於mxnet和gluoncv的圖像內容識別

程序環境爲高性能集羣：
CPU：Intel Xeon Gold 6140 Processor * 2（共36核心）
內存：512GB RAM
GPU：Tesla P100-PCIE-16GB * 2

In [1]:

%matplotlib inline

 

03. Predict with pre-trained YOLO models¶

This article shows how to play with pre-trained YOLO models with only a few lines of code.

First let's import some necessary libraries:

In [2]:

from gluoncv import model_zoo, data, utils
from matplotlib import pyplot as plt

 

Load a pretrained model¶

Let's get an YOLOv3 model trained with on Pascal VOC dataset with Darknet53 as the base model. By specifying pretrained=True, it will automatically download the model from the model zoo if necessary. For more pretrained models, please refer to :doc:../../model_zoo/index.

In [3]:

net = model_zoo.get_model('yolo3_darknet53_voc', pretrained=True)

 

Pre-process an image¶

Next we download an image, and pre-process with preset data transforms. Here we specify that we resize the short edge of the image to 512 px. You can feed an arbitrarily sized image. Once constraint for YOLO is that input height and width can be divided by 32.

You can provide a list of image file names, such as [im_fname1, im_fname2, ...] to :py:func:gluoncv.data.transforms.presets.yolo.load_test if you want to load multiple image together.

This function returns two results. The first is a NDArray with shape (batch_size, RGB_channels, height, width). It can be fed into the model directly. The second one contains the images in numpy format to easy to be plotted. Since we only loaded a single image, the first dimension of x is 1.

In [4]:

im_fname = utils.download('https://raw.githubusercontent.com/zhreshold/' +
                          'mxnet-ssd/master/data/demo/dog.jpg',
                          path='dog.jpg')
x, img = data.transforms.presets.yolo.load_test(im_fname, short=512)
#x, img = data.transforms.presets.yolo.load_test(im_fname)
print('Shape of pre-processed image:', x.shape)

 

Shape of pre-processed image: (1, 3, 512, 683)

 

Inference and display¶

The forward function will return all detected bounding boxes, and the corresponding predicted class IDs and confidence scores. Their shapes are (batch_size, num_bboxes, 1), (batch_size, num_bboxes, 1), and (batch_size, num_bboxes, 4), respectively.

We can use :py:func:gluoncv.utils.viz.plot_bbox to visualize the results. We slice the results for the first image and feed them into plot_bbox:

In [5]:

class_IDs, scores, bounding_boxs = net(x)

ax = utils.viz.plot_bbox(img, bounding_boxs[0], scores[0],
                         class_IDs[0], class_names=net.classes)
plt.show()

 

 

More Examples¶

In [12]:

files=["1.jpg","2.jpg","3.jpg","dining.jpg","nju-cs.jpg",  "nju-street.jpg",  "people.jpg",  "train.jpg",  "tree.jpg"]
for f in files:
    im_fname = utils.download('',path="data/"+f)
    x, img = data.transforms.presets.yolo.load_test(im_fname, short=2000)
    #x, img = data.transforms.presets.yolo.load_test(im_fname)
    print('Shape of pre-processed image:', x.shape)

    plt.imshow(img)

    class_IDs, scores, bounding_boxs = net(x)

    ax = utils.viz.plot_bbox(img, bounding_boxs[0], scores[0],
                             class_IDs[0], class_names=net.classes)
    plt.show()

 

Shape of pre-processed image: (1, 3, 730, 1024)

 

 

 

Shape of pre-processed image: (1, 3, 684, 1024)

 

 

 

Shape of pre-processed image: (1, 3, 768, 1024)

 

 

 

Shape of pre-processed image: (1, 3, 1024, 768)

 

 

 

Shape of pre-processed image: (1, 3, 768, 1024)

 

 

 

Shape of pre-processed image: (1, 3, 768, 1024)

 

 

 

Shape of pre-processed image: (1, 3, 683, 1024)

 

 

 

Shape of pre-processed image: (1, 3, 768, 1024)

 

 

 

Shape of pre-processed image: (1, 3, 1024, 768)

 

 

In [ ]:

 

 

tz@croplab,hzau