fine-grained cars classification
This is a simple pytorch re-implementation of CVPR 2018 Learning a Discriminative Filter Bank Within a CNN for Fine-Grained Recognition.
Introduction:
The features are summarized blow:
- Use VGG16 as base Network.
- got best result Top1 94%
- Datasets 35042 images Train/ Test ( 60% / 40%)
Pipeline
- process_data
- YOLOV3
- color detector
- DFL-CNN
Usage
sh run.sh