gplast / DroNet_PyTorch

Inference of DroNet on PyTorch

This repository is based on pytorch-yolo-v3

A PyTorch implementation of a DroNet Object Detector

Requirements

1. Python 3.5
2. OpenCV
3. PyTorch 0.4

Installation

sudo -H pip3 install torch==0.4.0

Using PyTorch 0.3 will break the detector.

Running the detector

Clone, and cd into the repo directory.

DroNet_car

python3 video_demo.py --cfg cfg/DroNet_car.cfg --weights cfg/DroNet_car.weights --video Car_Parking.mov --classes 1

DroNetV3_car

python3 video_demo.py --cfg cfg/DroNetV3_car.cfg --weights cfg/DroNetV3_car.weights --video Car_Parking.mov --classes 2

DroNetV3_car

python3 video_demo.py --cfg cfg/DroNetV3_car.cfg --weights cfg/DroNetV3_car.weights --video Car_Crossroad.mp4 --classes 2 --reso 1024