One of the key creative aspects of an advertisement is choosing the image that will appear alongside the advertisement text. The advertisers aim is to select an image that will draw the attention of the users and will get them to click on it, while remaining relevant to the advertisement text (naturally, an image of a cute puppy next to an advertisement about insurance doesn’t make much sense) . Below are some examples of advertisements appearing in Taboola’s “Promoted Links” box. Notice that each advertisement contains both a title and an appealing image.
Examples of advertisements placed in Taboola’s “Promoted Links” box. Notice that each advertisement contains both a title and an appealing image.
Given an advertisement title (for example “15 healthy dishes you must try”), the advertiser has endless possibilities of choosing the image thumbnail to accompany it, clearly some more clickable than others. One can apply best practices in choosing the thumbnail, but manually searching for the best image (out of possibly thousands that fit a given title) is time consuming and impractical. Moreover, there is no clear way of quantifying how much a given image is related to a title and more importantly – how clickable the image is, compared to other options.
To Alleviate this problem, we developed a text to image search algorithm that given a proposed title, scans an image gallery to find the most suitable images and estimates their expected click through rate (a common marketing metric depicting the amount of user clicks per a fixed number of advertisement displays).
For example, here are the images returned by our algorithm for the query title “15 healthy dishes you must try” along with their predicted Click Through Ratio (CTR):
Examples of images returned by our algorithm for the title query “15 healthy dishes you must try” along with their predicted Click Through Ratio (CTR). Notice that the images nicely fit the semantics of title.
Gil Levi and Tal Hassner, Emotion Recognition in the Wild via Convolutional Neural Networks and Mapped Binary Patterns, Proc. ACM International Conference on Multimodal Interaction (ICMI), Seattle, Nov. 2015
At the recent DevCon conference I had the pleasure of giving an introductory talk to Deep Learning. A short theoretical overview is given following a technical deep dive on how to train deep networks with a few demos, practical examples and tips.
The notebook used in the demo is available here and the various deep networks and definition files used to run the demo are available here.
In the last few posts we mostly talked about binary image descriptorsand the previous post in this line of works described our very own LATCH descriptor  and presented an evaluation of various binary and floating point image descriptors. In the current post we will shift our attention to the field of Deep Learning and present our work on Age and Gender classification from face image using Deep Convolutional Neural Networks .
Example images from the AdienceFaces benchmark
Our method was presented in the following paper:
Gil Levi and Tal Hassner, Age and Gender Classification using Convolutional Neural Networks, IEEE Workshop on Analysis and Modeling of Faces and Gestures (AMFG), at the IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Boston, June 2015.
In this post I will explain how to add a simple rotation invariance mechanism to the BRIEF descriptor, I will present evaluation results showing the rotation invariant BRIEF significantly outperforms regular BRIEF where visual geometric changes are present and finally I will post a C++ implementation integrated into OpenCV3.
We’ll start by a visual example, displaying the correct matches between a pair of images of the same scene, taken from different angles – once with the original version of BRIEF (first image pair) and one with the proposed rotation invariant version of BRIEF (second image pair):
Correct matches when using the BRIEF descriptor
Correct matches when using the rotation invariant BRIEF descriptor
It can be seen that there are much more correct matches when using the proposed rotation invariant of the BRIEF descriptor.
I recently came across a simple and easy package that can be used to create 3D reconstruction of objects. I wanted to share it and give an easy and practical explanation on how one can create visually appealing 3D models by running a few simple commands, no coding needed. I must emphasize that for keeping it simple, this post will not focus on theory as did the last few posts on binary descriptors, but instead will give an easy and practical guide to 3D reconstruction. Just to give you a taste of what can be done with the package, here’s an example of a 3D reconstruction I made (yes, that me in there):
Here you can see an original image vs. a screenshot of the 3D model:
This is our fifth post in the series about binary descriptors and here we will talk about the FREAK descriptor. This is the last descriptor that we’ll talk about as the next and final post in the series will give a performance evaluation of the different binary descriptors. Just a remainder – we had an introduction to binary descriptors and posts about BRIEF, ORB and BRISK.
We’ll start by showing the following figure that shows an example of using BRISK to match between real world images with viewpoint change. Green lines are valid matches, red circles are detected keypoints.
BRISK descriptor – example of matching points using BRISK
Lately, I’ve been reading a lot about BOW (Bag of Words) models  and I thought it would be nice to write a short post on the subject. The post is based on the slides from Li Fei-Fei taken from ICCV 2005 course about object detection:
As the name implies, the concept of BOW is actually taken from text analysis. The idea is to represent a document as a “bag” of important keywords, without ordering of the words (that’s why it’s a called “bag of words”, instead of a list for example).
Illustration of Bag of words model in documents
In computer vision, the idea is similar. We represent an object as a bag of visual words – patches that described by a certain descriptor:
Since the next few posts will talk about binary descriptors, I thought it would be a good idea to post a short introduction to the subject of patch descriptors. The following post will talk about the motivation to patch descriptors, the common usage and highlight the Histogram of Oriented Gradients (HOG) based descriptors.
I think the best way to start is to consider one application of patch descriptors and to explain the common pipeline in their usage. Consider, for example, the application of image alignment: we would like to align two images of the same scene taken at slightly different viewpoints. One way of doing so is by applying the following steps:
Compute distinctive keypoints in both images (for example, corners).
Compare the keypoints between the two images to find matches.
Use the matches to find a general mapping between the images (for example, a homography).
Apply the mapping on the first image to align it to the second image.