Fast target recognition based on pytorch and fast RCNN

Faster RCNN, be relative to R-CNN On the structure ,Faster RCNN Feature extraction has been done (feature extraction),proposal extract ,bounding box regression(rect refine),classification It's all integrated into one network , So that the comprehensive performance has been greatly improved , Especially in terms of detection speed .

PyTorch It's an open source Python Machine learning library , be based on Torch, For natural language processing applications . be relative to TensorFlow More lightweight , It is more suitable for scientific research and small projects .

Here is a simple example , be based on Fast RCNN Algorithm and PyTorch Fast target recognition . What we use here is coco Data sets have trained online models , It is relatively simple to operate .

Code

from PIL import Image
import matplotlib.pyplot as plt
# pip install -U matplotlib
import torch
# pip install pytorch
import torchvision.transforms as T
import torchvision
# pip install torchvision
import numpy as np
import cv2

import os

os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"

# pip install opencv-python



#  Download the trained model 
model = torchvision.models.detection.fasterrcnn_resnet50_fpn(pretrained=True)
model.eval()

COCO_INSTANCE_CATEGORY_NAMES = [
    '__background__', 'person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus',
    'train', 'truck', 'boat', 'traffic light', 'fire hydrant', 'N/A', 'stop sign',
    'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep', 'cow',
    'elephant', 'bear', 'zebra', 'giraffe', 'N/A', 'backpack', 'umbrella', 'N/A', 'N/A',
    'handbag', 'tie', 'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball',
    'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard', 'tennis racket',
    'bottle', 'N/A', 'wine glass', 'cup', 'fork', 'knife', 'spoon', 'bowl',
    'banana', 'apple', 'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza',
    'donut', 'cake', 'chair', 'couch', 'potted plant', 'bed', 'N/A', 'dining table',
    'N/A', 'N/A', 'toilet', 'N/A', 'tv', 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone',
    'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'N/A', 'book',
    'clock', 'vase', 'scissors', 'teddy bear', 'hair drier', 'toothbrush'
]


def get_prediction(img_path, threshold):
    img = Image.open(img_path)
    #  Change one PIL Library pictures or numpy The array of is tensor Tensor type ; Convert from [0,255]->[0,1]
    transform = T.Compose([T.ToTensor()])
    img = transform(img)

    pred = model([img])
    print(pred[0]['labels'].numpy())
    #  Category extraction 
    pred_class = [COCO_INSTANCE_CATEGORY_NAMES[i] for i in list(pred[0]['labels'].numpy())]
    #  Coordinate extraction 
    pred_boxes = [[(i[0], i[1]), (i[2], i[3])] for i in list(pred[0]['boxes'].detach().numpy())]

    #  Find the ones that meet the similarity requirements 
    pred_score = list(pred[0]['scores'].detach().numpy())
    pred_t = [pred_score.index(x) for x in pred_score if x > threshold][-1]
    pred_boxes = pred_boxes[:pred_t + 1]
    pred_class = pred_class[:pred_t + 1]
    print("pred_class:", pred_class)
    print("pred_boxes:", pred_boxes)
    return pred_boxes, pred_class


def object_detection_api(img_path, threshold=0.5, rect_th=3, text_size=3, text_th=3):
    boxes, pred_cls = get_prediction(img_path, threshold)
    img = cv2.imread(img_path)
    #  Convert to RGB Images 
    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    for i in range(len(boxes)):

        #  Circle the target according to the coordinates 
        cv2.rectangle(img, (int(boxes[i][0][0]), int(boxes[i][0][1])), (int(boxes[i][1][0]), int(boxes[i][1][1])),
                      color=(0, 255, 0),
                      thickness=rect_th)
        #  Label categories 
        cv2.putText(img, pred_cls[i], (int(boxes[i][0][0]), int(boxes[i][0][1])), cv2.FONT_HERSHEY_SIMPLEX, text_size,
                    (0, 255, 0), thickness=text_th)
    plt.imshow(img)
    plt.show()



if __name__ == '__main__':
    object_detection_api(img_path=r"C:\Users\hanbo\Pictures\dog.jpg")

result

Example 1

Example 2