Go to about page
Day Planner
- Model ensembling
- Make a classifier to filter out the FPs
- Try to figure out doing NMS with multiple image size inference like this
- Train Yolox, yolor
- Fix systamatic eval of yolov5
- Augment some data for validation
[Major ramp ups]
- Fine tune the provided models [IMP] [today]
- Train default CascadeRCNN with mmdet [IMP] [today]
- Use yolov5 evolution for HPT [IMP]
- Train yolox and yolor [IMP]
- Look into ensemble of hybrid model [IMP]
- Do the classification [IMP] [today]
- Submit the vid split yolov5 model [Minor] [today]
- Integrate the classification filter [IMP] [today]
- Systematic YOLO eval [IMP] [today]