This is a machine learning project which leverage RCNN model to detect multiple objects in an image. Specifically, the model will receive an image and spot out cars on the image.
The RCNN model follow the idea of these steps:
- Searching for regions that may have objects (called region proposal)
- Crop each region as a separate image
- Feed those images to a CNN model
Model architecture
For more detail, you can checkout ours explanation videos:
Regardings the pre-trained model, we have splitted it into 10 parts in models
. Please first use a joiner tool (example pinetool) to join those file into best_linear_svm_alexnet_car.pth
.
Then, you need to install all the requirement libraries:
pip install -r requirements.txt
Subsequently, using streamlit to host our website:
streamlit run streamlit_app.py
Our research team including:
This repository is built based on R-CNN repository by zhujian.
Reference papers:
@misc{girshick2013rich,
title={Rich feature hierarchies for accurate object detection and semantic segmentation},
author={Ross Girshick and Jeff Donahue and Trevor Darrell and Jitendra Malik},
year={2013},
eprint={1311.2524},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
@misc{pascal-voc-2007,
author = "Everingham, M. and Van~Gool, L. and Williams, C. K. I. and Winn, J. and Zisserman, A.",
title = "The {PASCAL} {V}isual {O}bject {C}lasses {C}hallenge 2007 {(VOC2007)} {R}esults",
howpublished = "http://www.pascal-network.org/challenges/VOC/voc2007/workshop/index.html"
}