Amirgu / MASH-Object-Recognition

MASH-Object-Recognition

School projects in the Object Recognition and Computer Vision Course (I. LAPTEV, J. PONCE, C. SCHMID, J. SIVIC) shared between Master 2 MASH in Paris Dauphine University and Master 2 MVA in ENS

Kaggle Competition: Bird Classification

MVA Kaggle Competition : Bird Classification (https://www.kaggle.com/competitions/mva-recvis-2022), Our solution got 84% Accuracy on the final test set. Download the training/validation/test images from here. The test image labels are not provided.

Cropping

Run the file crop.py

Data augmentation

Run The file augementation.py

Training

Run main.py

Evalution

Run eval.py

Object Detection and Tracking : DiffusionDet

We studied DiffusionDet which is a diffusion model for object detection. Specifically, we reproduced the results of the author on MS-COCO dataset and compared DiffusionDet performances with Faster R-CNN. The main goal was to extend the use of Diffusion model to Multi-Object Tracking. We implemented a centroid-based tracker on top of the DiffusionDet model.

Results

Compressed.mp4

Datasets

We have used the MS-COCO dataset for our object detection experiments and the MOT17 dataset for our Multi-Object Tracking experiments.

Bibliography

Chen & al., DiffusionDet : Diffusion Model for Object Detection (arxiv)

@misc{https://doi.org/10.48550/arxiv.2211.09788,
  doi = {10.48550/ARXIV.2211.09788},
  
  url = {https://arxiv.org/abs/2211.09788},
  
  author = {Chen, Shoufa and Sun, Peize and Song, Yibing and Luo, Ping},
  
  keywords = {Computer Vision and Pattern Recognition (cs.CV), FOS: Computer and information sciences, FOS: Computer and information sciences},
  
  title = {DiffusionDet: Diffusion Model for Object Detection},
  
  publisher = {arXiv},
  
  year = {2022},
  
  copyright = {Creative Commons Attribution 4.0 International}
}