Choosing the right tool

Will deep learning displace all other CV techniques?

It’s remarkable to see the range of applications in which deep neural networks are proving effective – often, significantly more effective than previously known techniques. From speech recognition to ranking web search results to object recognition, each day brings a new product or paper with a new challenge tamed by deep learning.

Artificial neural networks are a subset of machine learning techniques, and deep neural networks are a subset of artificial neural networks. All of these machine learning techniques are used in computer vision - and for other fields. (Bild: Embedded Vision Alliance)

Artificial neural networks are a subset of machine learning techniques, and deep neural networks are a subset of artificial neural networks. All of these machine learning techniques are used in computer vision – and for other fields. (Bild: Embedded Vision Alliance)

Anzeige

Computer vision, of course, is a field with significant deep learning activity. Deep learning is particularly appealing for visual perception because image and video data is massive and rich with information, but is infinitely variable and often ambiguous. Researchers have demonstrated deep neural networks capable of classifying skin lesions as cancerous or benign with accuracy equivalent to that of dermatologists, and other deep networks are capable of reading lips with significantly higher accuracy than human lip readers. More recently, Shunsuke Saito, Lingyu Wei and colleagues have developed a deep-learning-based approach that generates photorealistic 3D models of human faces from a single 2D image. This naturally leads to the question: Will deep learning displace all other techniques for solving visual perception problems? While the power of deep learning is undeniable, I believe that other techniques will continue to be attractive. As in most fields, it’s a question of choosing the right tool for the job. With the current justifiable excitement around deep learning, one important point is that machine learning encompasses a vast range of techniques, of which neural networks are a subset (as shown in the figure). In turn, deep neural networks are a subset of a broader class of neural network techniques. In many cases, simpler machine learning techniques are effective. That’s fortunate, because deep learning has a deep appetite: massive amounts of data and compute power are needed for training deep networks, and substantial compute power is needed for running trained networks. Other machine learning techniques often get by with comparatively minuscule amounts of data and processing power. Well before the start of the current deep learning craze, a few pioneering researchers and companies, such as BrainChip and General Vision, were deploying what we might today call ’shallow learning‘ algorithms and processors to solve real-world problems ranging from fish identification to license-plate recognition. Today, there’s a huge rush to apply deep learning to every visual perception problem. But, particularly when development costs and deployment costs are considered, I think that in many cases, other techniques will prove preferable. The challenge for algorithm, system, and product developers will be how to discern which tasks call for deep learning, which call for other neural network techniques, which are best served by other machine learning approaches, and which are still appropriate for good old-fashioned hand-crafted algorithms. On September 7th there will be a full day course in Hamburg which provides the hands-on knowledge you need to develop deep learning computer vision applications with TensorFlow. For details see https://tensorflow.embedded-vision.com.

Choosing the right tool
Bild: Embedded Vision Alliance


Das könnte Sie auch interessieren

www.baslerweb.com

Anzeige

Basler und Arrow Electronics haben ein Vertriebsabkommen unterzeichnet, laut dem Arrow die Produktpalette von Basler ab dem 1. März in der EMEA-Region vertreiben wird. Darüber hinaus hat Arrow die Produkte auch in seine digitale globale Linecard aufgenommen. Mit einem besonderen Augenmerk auf die Embedded Vision Produkte von Basler arbeitet Arrow daran, diese Technologie in sein eigenes SoM- und Boards-Portfolio zu integrieren.

www.baslerweb.com

Anzeige

Laut einer aktuellen Studie von Yole soll sich der Markt für industrielle Bildverarbeitungskameras von 2Mrd.USD (2017) auf 4Mrd.USD bis 2023 verdoppeln. Wachstumstreiber für Machine Vision seien dabei vor allem nicht-industrielle Anwendungen und auch der Markt für autonome Fahrzeuge.

www.i-micronews.com

Anzeige

Seit dem 1. Februar ist die deutsche Niederlassung von Edmund Optics in Mainz in der Isaac-Fulda-Allee 5 zu finden. Durch den Umzug agiert Edmund Optics mit der zu Beginn des Jahres akquirierten Itos GmbH nun gestärkt vom Standort Mainz.

www.edmundoptics.de

Anzeige

Lucid Vision Labs hat die derzeit weltweit kleinste GigE Vision PoE-Kamera namens Phoenix entwickelt. Der Kunde erhält mit diesem Modell ein flexibles Kameradesign zu einem angemessenen Preis-/Leistungsverhältnis. Zusätzlich zeichnet sich die Kamera durch ein kompaktes NF-Mount 5MP Objektiv sowie optional den neuen ix Industrial Ethernet Anschlüssen aus.‣ weiterlesen

www.thinklucid.com

In Zusammenarbeit mit dem CDTM-Institut der TU München hat Framos ein Wearable entwickelt, das mit Hilfe von 3D-Echtzeit-Technologie Sehbehinderte im Alltag unterstützt. Die Brillen sind mit Intel RealSense Stereokameras ausgestattet, wobei intelligente Algorithmen die visuellen Signale in haptische und akustische Informationen übersetzen. Die Audioinformation basiert auf der Erkennung von Objekten und Schriften, die haptische Rückmeldung über Positionen und Distanzen erhalten die Sehbehinderten über ein mit Vibrationsmotoren ausgestattetes Armband.

www.framos.com

Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige