A real-time object detection app based on lightDenseYOLO Our lightDenseYOLO is the combination of two components: lightDenseNet as the CNN feature extractor and YOLO v2 as the detection module
A pytorch implementation of YOLOv1-v3.
Project only supports python3.x.
To do
Test the model of YOLOv1.
Train the model of YOLOv1.
Test the model of YOLOv2.
Train the model of YOLOv2.
Test the model of YOLOv3.
Train the model of YOLOv3.
Data augmentation.
Using k-means to generate the priors.
Evaluate the model including mAP, precision, recall, etc.
Data of VOC format -> YOLO format.
Dependency
torch 0.3.1
opencv-python
torchvision
numpy
pillow
argparse
Train
Prepare
VOC -> YOLO
Data of VOC format(lxml) -> YOLO format(txt).
You can use the script in ./TOOL/voc2yolo.py to complete the work of conversion.
Usage
modify Line<13~15> according to your needs
run "python3 voc2yolo.py"
Get good priors
Run k-means clustering on the dimensions of bounding boxes to get good priors for our model.
You can use the script in ./TOOL/genPriors/genPriors.py to get the good priors.
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].