Object Detection Notifier
Detect objects, capture it and send them to mail with the image attachment
Follow the below commands step by step to make this project run
Clone this repository and execute the commands as shown below
git clone https://github.com/VigneshKathirkamar/object_detection_notifier.git
cd object_detection_notifier
wget "http://download.tensorflow.org/models/object_detection/ssd_mobilenet_v2_coco_2018_03_29.tar.gz"
Once the tar file gets downloaded, unzip it in the same folder (object_detection_notifier)
pip3 install -r requirements.txt
Generation of 16 digit password for you mail id
Follow this link:https://qr.ae/pGepQ7 to generate 16 digit password for your mail id
Run the below command with your user name and password.
python3 person_detector_video.py --to_mail_id karthik.akash08@gmail.com --pwd qlaxxlvqpknkhhqp --from_mail_id i51vignesh@gmail.com
To do
- Scaling the project. Right now, since we are using gmail, the total number of notifications/mails is limited to 2000 per person per day.(i.e 500 mails for 4 person per day)
- Instead of person detection, models like MTCNN can be used to crop the face alone and this data can be used for training a classifier model. This would find use case in retail store analysis
Update
Tried using Raspberry pi (pi 3 Model B+) by cloning the repo, everything works fine but the latency time is very slow that it can't be used for real time puproses. Camera used was pi 5MP camera board module [https://www.amazon.in/gp/product/B00E1GGE40/ref=ppx_yo_dt_b_asin_image_o00_s00?ie=UTF8&psc=1]
GPU has to be used for reducing the latency time