The other day, at a wedding reception of an acquaintance, I was asked to do entertainment for the first time, and I managed to survive, so that was the time.
I don't have anything to have. I'm not a master of singing or dancing. I can make some movies, but it seems that I will do it for the sideshow of the bride. .. : There is only programming! !!
** "HAPPINESS SCOUTER" **
When you send a photo on LINE, it recognizes the facial expression of the person in the picture and measures ** "happy combat power" **! !! something like. (This is an extension of LINE BOT using Amazon Rekognition, which I wrote in Previous article) I also created a page where you can see the list and details of the photos you sent. On the list page, it is displayed in the ranking format of happy combat power.
[LINE screen]
[Detailed page]
[List page]
--QR code for LINE account and list page
We prepared business card-sized cards for each person and handed them out to each table between the wedding and reception on the day.
Design: Canva Printing: Business card shop dot com → Supports submission in png, ships on the same day, and receives 500pt for the first time, so it was cheap at 660 yen for 100 sheets!
--Explanatory slides
Created with Google Slides. Actually, I also include a brief self-introduction at the beginning. Pakutaso's photo is very helpful lol
--Premium
The 1st place was given a LINE prepaid card worth 1000 yen.
--Simple self-introduction and explanation while projecting slides on the screen. ――Measure the time for 5 minutes and have the photograph taken on the spot measured. A list page is displayed on the screen to show the actual situation. ――Finally, check the result and give a prize to the person who took the picture with the highest fighting power. ――Be free to use after entertainment. (During the second party)
I decided to participate freely because I need to make friends with my LINE account. 48 out of 66 guests participated.
We were able to collect about 80 photos in 5 minutes of sideshow, and finally 150 photos. I also collected unrelated photos, but ... lol
I used Python as the language and Amazon Rekognition for the face recognition part, and configured it as follows. I'm imitating various things. ..
The Flask framework is used for the list / detail page, and Masonry.js is used for arranging the photos on the list page.
--How to score In many cases, the results were negative because emotions other than "happiness" and "calmness" were calculated negatively. It was a reception, so I should have thought about it a little more. ..
--The accuracy of the target face As shown below, even the people in the back who have nothing to do with it were recognized and affected the score. Can you target by the size of the bounding box?
Try using the image processing library "Pillow" with AWS Lambda Display Boundary Box I tried DynamoDB using Python (boto3)
If you understand how to use Jinja2, development using Flask will become smarter The strongest compatibility between Bootstrap 4 Cards and Masonry.js
Recommended Posts