Intelligent pest control application - CNN and water turret to defend bird feeder from squirrels
Tired of having squirrels eat all your bird seed? Squirtl protects your birdfeeder by identifying, tracking and accurately spraying squirrels with a burst of water! Squirtl is non-lethal, and can be retrained to apply to numerous applications. A simplified version of the targeting algorithm is included with the repository.
A video demonstration and further information can be found at www.squirtl.com
- Install awscli
sudo pip3 install awscli
- Get the relevant OpenImages files needed to locate images of our interest
wget https://storage.googleapis.com/openimages/2018_04/class-descriptions-boxable.csv
wget https://storage.googleapis.com/openimages/2018_04/train/train-annotations-bbox.csv
- Download the images from OpenImagesV4
python3 getDataFromOpenImages_squirrel.py
- Create the train-test split
python3 splitTrainAndTest.py /home/naivoder/feeder/CNN/JPEGImages
Give the correct path to the data JPEGImages folder. The 'labels' folder should be in the same directory as the JPEGImages folder.
- Install Darknet and compile it.
cd ~
git clone https://github.com/pjreddie/darknet
cd darknet
make
- Get the pretrained model
wget https://pjreddie.com/media/files/darknet53.conv.74 -O ~/darknet/darknet53.conv.74
-
Fill in correct paths in the darknet.data file
-
Start the training as below, by giving the correct paths to all the files being used as arguments
cd ~/darknet
./darknet detector train /home/naivoder/feeder/CNN/darknet.data /home/naivoder/feeder/CNN/darknet-yolov3.cfg ./darknet53.conv.74 > /home/naivoder/feeder/CNN/train.log
- Give the correct path to the modelConfiguration and modelWeights files in object_detection_yolo.py and test any image or video for squirrel detection, e.g.
python3 object_detection_yolo.py --image=squirrelImage.jpg
python3 object_detection_yolo.py --video=squirrelVideo.mp4
python3 object_detection_yolo.py *defaults to webcam