I implemented and practice with object recognition. This assignment has two parts. In Part 1, I implemented the basic object recognition+localization pipeline. In Part 2, I used YOLOv3 object detection with mean-shift algorithm for tracking.
from PIL import Image import torch import torchvision from matplotlib import patches import numpy as np import matplotlib.pyplot as plt
import torch.nn as nn import torch.optim as optim import torch.nn.functional as F import time import cv2 import csv
The dataset folder must be in the same directory as the python files or the path can be changed from the config.py file. At the same time, the variables can be changed from the config.py file. Part1: Train, aval and test files can be run internally. Part2: It works with the main file.