6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 ·...
Transcript of 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 ·...
![Page 1: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/1.jpg)
6.S097 Final PresentationMoin Nadeem, Ramya Durvasula, Faraaz Nadeem,
Amy Fang
![Page 2: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/2.jpg)
Problem- We want to be able to guess house prices- Most house price estimates do not incorporate visual aspects, but there is
obviously some correlation between how a house looks and its cost- This problem hasn’t been previously solved: no one uses image in price
prediction- Hence, no dataset exists.
![Page 3: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/3.jpg)
Our Solution- Wrote a scraper to take data from Zillow, and trained a neural network to take
as input a picture of a house and return the estimated price range - Only used the raw pixel data of houses, no data about square footage,
number of beds/baths, last sale price, relative location, etc- Augmented the training data to prevent overfitting (adding pictures with
distortion, rotation, translation, reflection, etc.)- 8 “buckets” of price ranges
![Page 4: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/4.jpg)
train_datagen = ImageDataGenerator( rescale=1./255, shear_range=0.2, zoom_range=0.2, horizontal_flip=True)
![Page 5: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/5.jpg)
![Page 6: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/6.jpg)
Model Architecture
![Page 7: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/7.jpg)
Model Architecture- Batch Size = 32, 16- 32x32 Filter Size- 2x2 Max Pooling- Buckets:
- 0 to 100k- 100k to 200k- 200k to 350k- 350k to 550k- 550k to 800k- 800k-1.1m- 1.1-1.45m- 1.45m-3m
![Page 8: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/8.jpg)
Visualization
(all convolution filters available at http://moinnadeem.com/housing_prediction/conv_1/)
![Page 9: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/9.jpg)
Visualization… oh!
(all convolution filters available at http://moinnadeem.com/housing_prediction/conv_1/)
![Page 10: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/10.jpg)
Visualization, done properly.
(all convolution filters available at http://moinnadeem.com/housing_prediction/conv_1/)
![Page 11: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/11.jpg)
Visualization
![Page 12: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/12.jpg)
Performance- Model performed at 40% accuracy on the testing set- Given that we had 8 buckets, the accuracy of a random model would have
been 12.5%, so this is a decent sized improvement- Improvements
- Adam optimizer over Adadelta / RMSprop- “Momentum” in Adam proved to be pivotal
- Changed batch size from 32 to 16- Also solved memory issues
- Used an ImageDataGenerator to skew in order to prevent overfitting.
- We may be able to substantially improve the accuracy if we added more data (currently at 3k)
![Page 13: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/13.jpg)
Learned Lessons- Data is everything- Use more epochs- What if we added zip codes or did
parameter tuning?
![Page 14: 6.S097 Final Presentation - MITweb.mit.edu/6.S097/www/projects/CVRealEstate.pdf · 2017-02-02 · Moin Nadeem, Ramya Durvasula, Faraaz Nadeem, Amy Fang. Problem - We want to be able](https://reader033.fdocuments.in/reader033/viewer/2022050422/5f916c3906ea2f53ce215b83/html5/thumbnails/14.jpg)
Future Improvements- Add relative location features such
as zipcode, lat/long values- Merge layer
- Add overhead view, indoor view, other standard features such as number of bed/baths, square footage, etc.