The document lists several formats for exporting a YOLOv5s object detection model, including PyTorch (.pt), TorchScript (.torchscript), ONNX (.onnx), OpenVINO (directory), TensorRT (.engine), TensorFlow SavedModel (directory), TensorFlow JS (.json), and TensorFlow Lite (.tflite). It also provides a short Python code example for loading the PyTorch model and performing inference on images.

6.  Format  export.py --include  Model
 PyTorch  custom  yolov5s.pt
 TorchScript  torchscript  yolov5s.torchscript
 ONNX  onnx  yolov5s.onnx
 OpenVINO  openvino  yolov5s_openvino_model/
 TensorRT  engine  yolov5s.engine
 TensorFlow
SavedModel
 saved_model  yolov5s_saved_model/
 TensorFlow JS  json  yolov5s.json
 TensorFlow Lite  tflite  yolov5s.tflite

8. PyTorch

15. Import cv2
import torch
Import numpy as np
# Model
model = torch.hub.load('ultralytics/yolov5', 'yolov5s', pretrained=True)
model = torch.hub.load('ultralytics/yolov5’, ‘custom’, path=’best.pt')
# Images
imgs = ['https://ultralytics.com/images/zidane.jpg'] # batch of images
# Inference
results = model(imgs)
# Results
results.print()
results.save()
results.show()
results.xyxy[0] # img1 predictions (tensor)
results.pandas().xyxy[0] # img1 predictions (pandas)

16. Object detection is a computer vision task that involves identifying and localizing objects
in an image or video. It is a challenging task that requires both localizing objects in the
image and classifying what type of object has been detected.
There are primarily two types of deep learning-based object detection algorithms: one-
stage and two-stage detectors.
One-stage detectors, such as YOLO, SSD, and RetinaNet, directly predict bounding
boxes and class probabilities in a single pass. These algorithms are much faster than
two-stage detectors, but they are less accurate.
Two-stage detectors, such as RCNN, Faster R-CNN, and Mask R-CNN, use two
networks: The first network, called a region proposal network (RPN), proposes regions
where objects are likely to be present. The second network (a classifier) classifies the
proposed regions into different classes of objects