1
$\begingroup$

I have a relatively simple object localization task. I have an image set that are either uniform or contain an 'object' consisting of a black circular shape at some position in the image. The labels include whether or not it contains the object, and the position of the object (center coordinates + length/width). YOLO feels like overkill and I'm having a hard time figuring out how to get it to work, but I'm having a hard time finding a simpler model that would work.

$\endgroup$
1
  • $\begingroup$ What is learning bringing to this party? Machine vision techniques would be quite simple compared to learning. $\endgroup$
    – Stephen Rauch
    Commented Apr 19, 2019 at 19:15

1 Answer 1

1
$\begingroup$

So, the following techniques are used for object detection (other can YOLO, R-CNN and other Deep Learning Techniques):

Haar Cascade Classifier

This was the first facial detection algorithm to be successfully used and is popular until today. Proposed by Viola and Jones in their paper Rapid Object Detection using a Boosted Cascade of Simple Features (2001, CVPR) this algorithm uses Haar-like features from integral images and a cascade of boosting classifiers to detect objects in images. Lienhart proposed in 2002 a improvement to this algorithm in his paper Empirical Analysis of Detection Cascades of Boosted Classifiers for Rapid Object Detection

It can be easily trained using OpenCV Object Detection Module in Python, C and Java, also available in Matlab Computer Vision Tool Box as vision.CascadeObjectDetector and probably many other platforms.

Max-Margin Object Detection

Proposed in the paper Max-Margin Object Detection by Davis E. King, creator of dlib, where it has been implemented, this algorithm has implementations for face detection using HoG Features and CNNs features.

The abstract of the paper goes:

Most object detection methods operate by applying a binary classifier to sub-windows of an image, followed by a non-maximum suppression step where detections on overlapping sub-windows are removed. Since the number of possible sub-windows in even moderately sized image datasets is extremely large, the classifier is typically learned from only a subset of the windows. This avoids the computational difficulty of dealing with the entire set of subwindows, however, as we will show in this paper, it leads to sub-optimal detector performance.

In particular, the main contribution of this paper is the introduction of a new method, Max-Margin Object Detection (MMOD), for learning to detect objects in images. This method does not perform any sub-sampling, but instead optimizes over all sub-windows. MMOD can be used to improve any object detection method which is linear in the learned parameters, such as HOG or bag-of-visual-word models. Using this approach we show substantial performance gains on three publicly available datasets. Strikingly, we show that a single rigid HOG filter can outperform a state-of-the-art deformable part model on the Face Detection Data Set and Benchmark when the HOG filter is learned via MMOD.

I think these are the two most relevant outside of Deep Learning, reading the related work fromt he 3 cited articles might give you some light on how to proceed.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.