Automated Attendance System Using Deep Learning - Revision history

imported>Student2018: /* Appendix */

2018-12-16T20:05:38Z

Appendix

← Older revision		Revision as of 20:05, 16 December 2018
Line 123:		Line 123:
	[[File:BlurComparison.png\|center\|frame\|(From left to right) Average, Median and Gaussian blur, (from top to bottom) blur level 5 and 27.]]		[[File:BlurComparison.png\|center\|frame\|(From left to right) Average, Median and Gaussian blur, (from top to bottom) blur level 5 and 27.]]

			=== Code and Dataset ===
			To download the repository:

			git clone https://username@bitbucket.org/jpergament/psych221-project.git

			This repository contains a README file which describes the code and the used images.

	== References ==		== References ==

imported>Student2018: /* Results */

2018-12-15T06:17:36Z

Results

← Older revision		Revision as of 06:17, 15 December 2018
Line 47:		Line 47:
	[[File:Sharon2OriginalQuality.jpg\|center\|frame\|Original image (29KB). ]]		[[File:Sharon2OriginalQuality.jpg\|center\|frame\|Original image (29KB). ]]

	After resizing and compressing the image with a quality factor of 10 ~~and a size reduction of 86.2%~~:		After resizing and compressing the image with a quality factor of 10:

	[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 (4KB). ]]		[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 and a size reduction of 86.2% (4KB).]]

	We can see the effects that compression has on each individual pixel of image. Below is shown a comparison between a zoomed in area of size 24x24 pixels of the original image, and the same area of the compressed image with a quality factor of 10. This loss of information could have a significant impact on the ability of the inference model to make accurate predictions, since the information of relevant facial features and landmarks is compromised.		We can see the effects that compression has on each individual pixel of image. Below is shown a comparison between a zoomed in area of size 24x24 pixels of the original image, and the same area of the compressed image with a quality factor of 10. This loss of information could have a significant impact on the ability of the inference model to make accurate predictions, since the information of relevant facial features and landmarks is compromised.

imported>Student2018: /* Results */

2018-12-15T06:14:53Z

Results

← Older revision		Revision as of 06:14, 15 December 2018
Line 47:		Line 47:
	[[File:Sharon2OriginalQuality.jpg\|center\|frame\|Original image (29KB). ]]		[[File:Sharon2OriginalQuality.jpg\|center\|frame\|Original image (29KB). ]]

	After compressing the image with a quality factor of 10 and a size reduction of 86.2%:		After resizing and compressing the image with a quality factor of 10 and a size reduction of 86.2%:

	[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 (4KB). ]]		[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 (4KB). ]]

imported>Student2018 at 06:07, 15 December 2018

2018-12-15T06:07:24Z

← Older revision		Revision as of 06:07, 15 December 2018
Line 31:		Line 31:
	=== Data set and Inference model ===		=== Data set and Inference model ===

	For this project we used [http://vis-www.cs.umass.edu/lfw/ Labeled Faces in the Wild (LFW) data-set]. This data set contains more than 13000 images of 5749 different people [http://vis-www.cs.umass.edu/lfw/lfw.pdf]. 1680 people have two images or more, whereas the other 4069 people have only one image. In order to train an accurate CNN, one needs to have multiple images of the same person. Therefore, in our project we chose the 10 people in the LFW data set who have the most images~~. Each of the ten people~~, ~~have more than~~ 50 ~~different~~ images. In order to create more images, we used augmentation techniques such as shifting and rotation.		For this project we used [http://vis-www.cs.umass.edu/lfw/ Labeled Faces in the Wild (LFW) data-set]. This data set contains more than 13000 images of 5749 different people [http://vis-www.cs.umass.edu/lfw/lfw.pdf]. 1680 people have two images or more, whereas the other 4069 people have only one image. In order to train an accurate CNN, one needs to have multiple images of the same person. Therefore, in our project we chose the 10 people in the LFW data set who have the most images, over 50 images each. In order to create more images, we used augmentation techniques such as shifting and rotation.

	We used a pre-trained Inception V3 CNN model. We used transfer learning techniques and modified the last couple of layers in order to adjust to the recognition task. These layers were trained using the data-set specified above. Our code is based on [https://medium.com/@williamkoehrsen/facial-recognition-using-googles-convolutional-neural-network-5aa752b4240e?fbclid=IwAR17sCdzMhOs7b9ItQV2WBZ3xJOESm5xxOJyY_i8ukNffZnCHZ4hb1RkC2E this project]. We made slight modifications to achieve higher accuracy.		We used a pre-trained Inception V3 CNN model. We used transfer learning techniques and modified the last couple of layers in order to adjust to the recognition task. These layers were trained using the data-set specified above. Our code is based on [https://medium.com/@williamkoehrsen/facial-recognition-using-googles-convolutional-neural-network-5aa752b4240e?fbclid=IwAR17sCdzMhOs7b9ItQV2WBZ3xJOESm5xxOJyY_i8ukNffZnCHZ4hb1RkC2E this project]. We made slight modifications to achieve higher accuracy.

imported>Student2018: /* Results */

2018-12-15T03:47:14Z

Results

← Older revision		Revision as of 03:47, 15 December 2018
Line 54:		Line 54:

	<div><ul>		<div><ul>
	<li style="display: inline-block;"> [[File:SharonZoomOriginal.png\|thumb\|left\|Zooming in on a 24x24 pixel area the original image]] </li>		<li style="display: inline-block;"> [[File:SharonZoomOriginal.png\|thumb\|left\|Zooming in on a 24x24 pixel area the original image.]] </li>
	<li style="display: inline-block;"> [[File:SharonZoom10.png\|thumb\|right\|Zooming in on compressed image with quality factor of 10]] </li>		<li style="display: inline-block;"> [[File:SharonZoom10.png\|thumb\|right\|Zooming in on compressed image with quality factor of 10.]] </li>
	</ul></div>		</ul></div>

imported>Student2018: /* Results */

2018-12-15T03:46:51Z

Results

← Older revision		Revision as of 03:46, 15 December 2018
Line 51:		Line 51:
	[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 (4KB). ]]		[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 (4KB). ]]

	We can see the effects that ~~the~~ compression has on each individual pixel of ~~the images~~. Below is shown a comparison between a zoomed in area of the original image ~~of size 24x24 pixels~~, and the same area of the compressed image with a quality factor of 10.		We can see the effects that compression has on each individual pixel of image. Below is shown a comparison between a zoomed in area of size 24x24 pixels of the original image, and the same area of the compressed image with a quality factor of 10. This loss of information could have a significant impact on the ability of the inference model to make accurate predictions, since the information of relevant facial features and landmarks is compromised.

	<div><ul>		<div><ul>

imported>Student2018: /* Results */

2018-12-15T03:18:08Z

Results

← Older revision		Revision as of 03:18, 15 December 2018
Line 54:		Line 54:

	<div><ul>		<div><ul>
	<li style="display: inline-block;"> [[File:SharonZoomOriginal.png\|thumb\|left\|Zooming in on original image]] </li>		<li style="display: inline-block;"> [[File:SharonZoomOriginal.png\|thumb\|left\|Zooming in on a 24x24 pixel area the original image]] </li>
	<li style="display: inline-block;"> [[File:SharonZoom10.png\|thumb\|right\|Zooming in on compressed image with quality factor of 10]] </li>		<li style="display: inline-block;"> [[File:SharonZoom10.png\|thumb\|right\|Zooming in on compressed image with quality factor of 10]] </li>
	</ul></div>		</ul></div>

imported>Student2018: /* Results */

2018-12-15T03:16:02Z

Results

← Older revision		Revision as of 03:16, 15 December 2018
Line 51:		Line 51:
	[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 (4KB). ]]		[[File:Sharon2Quality10.jpg\|center\|frame\|Compressed image with a quality factor of 10 (4KB). ]]

	~~[[File:SharonZoomOriginal~~.~~jpg\|center\|frame\|600px]]~~		We can see the effects that the compression has on each individual pixel of the images. Below is shown a comparison between a zoomed in area of the original image of size 24x24 pixels, and the same area of the compressed image with a quality factor of 10.
	~~[[File:SharonZoom10~~.~~jpg\|center\|frame\|600px]]~~

			<div><ul>
			<li style="display: inline-block;"> [[File:SharonZoomOriginal.png\|thumb\|left\|Zooming in on original image]] </li>
			<li style="display: inline-block;"> [[File:SharonZoom10.png\|thumb\|right\|Zooming in on compressed image with quality factor of 10]] </li>
			</ul></div>

	After compression, we ran the images through the inference network to find the accuracy rates with different levels of compression to find a correlation between accuracy, file size reduction and quality factor.		After compression, we ran the images through the inference network to find the accuracy rates with different levels of compression to find a correlation between accuracy, file size reduction and quality factor.

	<div><ul>		<div><ul>
	<li style="display: inline-block;"> [[File:AccuracyVsQuality.png\|thumb\|left]] </li>		<li style="display: inline-block;"> [[File:AccuracyVsQuality.png\|thumb\|left\|600px]] </li>
	<li style="display: inline-block;"> [[File:SizeReducVsQuality.png\|thumb\|right]] </li>		<li style="display: inline-block;"> [[File:SizeReducVsQuality.png\|thumb\|right\|600px]] </li>
	</ul></div>		</ul></div>

imported>Student2018: /* Results */

2018-12-15T03:11:09Z

Results

← Older revision		Revision as of 03:11, 15 December 2018
Line 58:		Line 58:

	<div><ul>		<div><ul>
	<li style="display: inline-block;"> [[File:AccuracyVsQuality.~~jpg~~\|thumb\|left]] </li>		<li style="display: inline-block;"> [[File:AccuracyVsQuality.png\|thumb\|left]] </li>
	<li style="display: inline-block;"> [[File:SizeReducVsQuality.~~jpg~~\|thumb\|right]] </li>		<li style="display: inline-block;"> [[File:SizeReducVsQuality.png\|thumb\|right]] </li>
	</ul></div>		</ul></div>

imported>Student2018: /* Results */

2018-12-15T03:10:25Z

Results

← Older revision		Revision as of 03:10, 15 December 2018
Line 58:		Line 58:

	<div><ul>		<div><ul>
	<li style="display: inline-block;"> [[File:AccuracyVsQuality.~~png~~\|thumb\|left~~\|600px~~]] </li>		<li style="display: inline-block;"> [[File:AccuracyVsQuality.jpg\|thumb\|left]] </li>
	<li style="display: inline-block;"> [[File:SizeReducVsQuality.~~png~~\|thumb\|right~~\|600px~~]] </li>		<li style="display: inline-block;"> [[File:SizeReducVsQuality.jpg\|thumb\|right]] </li>
	</ul></div>		</ul></div>