Tensorflow object recognition

I am doing a project for a local non-profit museum and due to the sensitivity of the artifacts displayed, visitors are supposed to store all their cameras and cell phones at the counter before entering the museum. However, many visitors need to be reminded and one possible solution is to use Tensorflow object recognition technology to produce beeping sounds whenever our installed camera detects someone entering the museum with a camera or cell phone. I am using Raspberry Pi 3 (a single board computer with Graphics Processing Unit (GPU) and Central Processing Unit (CPU). Here are the specs in detail: [login to view URL] The Operating System (OS) is running on Linux/Debian and I have both Python 3 and Tensorflow installed. I have also connected a 5mp camera attached: https://www.amazon.com/Arducam-Megapixels-Sensor-OV5647-Raspberry/dp/B012V1HEP4/ref=sr_1_2?s=electronics&ie=UTF8&qid=1473286363&sr=1-2&keywords=raspberry+pi+camera And I have also connected a speaker to one of the four USB ports available on Raspberry Pi 3. You may use Putty to remotely access my Raspberry Pi 3 using the IP address via SSH. I will also provide you with the login details. 1. TensorFlow comes with a prebuilt model called “inception” that performs object recognition. Here's the tutorial to get it running: [login to view URL] 2. I need you to add or modify the codes so that whenever "cellphone" or "camera" is recognized, (a) it will send a signal to the speaker to produce 3 beeps sounds... Beep-beep-beep! (b) it will also record a X mili seconds video and save it to a specific path. When video recording is taking place, no signals will be sent to the speaker to make beeping sounds. Sending signals to the speaker will resume after the video recording has ended. 3. I should be able to able to configure the following: (a) the length of the recorded video, X, in mili seconds (b) Keywords to be detected. Currently, only "camera" and cellphone", but I may add more keywords. (c) I should be able to specify the path of the recorded video 4. You need to have experience in Tensorflow and some experience in Raspberry Pi 3 before bidding for this project. This project is inspired by and you can find relevant info there. [login to view URL] Thank you!