# Tensorflow GPU
conda env create -f conda-gpu.yml
conda activate yolov4-gpu
Make sure to use CUDA Toolkit version 10.1 as it is the proper version for the TensorFlow version used in this repository. https://developer.nvidia.com/cuda-10.1-download-archive-update2
USE MY FINAL CUSTOM TRAINED CUSTOM WEIGHTS: https://drive.google.com/file/d/1EYKcptLtDeLJJNhYKUpyOKqWdzhkopk4/view?usp=sharing
Copy and paste customfinals .weights file into the 'data' folder
To implement YOLOv4 using TensorFlow, first we convert the .weights into the corresponding TensorFlow model files and then run the model.
# Convert darknet weights to tensorflow
## custom
python save_model.py --weights ./data/customfinals.weights --output ./checkpoints/customfinals-416 --input_size 416 --model yolov4
# Run yolov4 tensorflow model
python main.py
Huge shoutout goes to hunglc007 for creating the backbone of this repository: