Classification and object detection are two impor�tant problems to be solved for developing automatic systems. The evolution of machine learning, especially deep learning, has showcased astounding success in performing these tasks. With that inspiration, this work performs a classification task on a flower species dataset and an object detection task on car detection dataset with the help of deep neural networks. The neural networks are built up from scratch and the concept of image processing has also been applied to perform these tasks.
Due to huge size, all the dataset and models are uploaded to following link: https://uflorida-my.sharepoint.com/:f:/g/personal/dsaha_ufl_edu/EkchgG6JAFBMj1xSlt6wNNoBWk_9G_ClsGQoiN178C9xVA?e=4nMj1b
This link contains two folders named 'problem1' and 'problem2' for Flower Species Classification and Object Detection Dataset, respectively.
How to run training file: Collect the models from problem1/final_model and problem1/other_models download training.ipyb from repository run the training.ipynb file
How to test: Collect the models from problem1/final_model download test.ipyb from repository run test.ipynb file
A csv for test set has been created and the csv for training has modified by including no car images The csv files can be found in problem2/data HOG images have been used. The dataset with hog images can be found in problem2/data normalized labeling will be also required. These labels can be found in problem2/data
How to run training file:
Collect the models from problem2/final_model and problem1/other_models Collect training csv file from problem2/data download training.ipyb from repository run the training.ipynb file
How to test: Collect the models from problem2/final_model Collect test csv file from problem2/data download test.ipyb from repository run test.ipynb file