Performance Evaluation of Neuromorphic-Vision Object Recognition Algorithms

Kasturi, Rangachar; Goldgof, Dmitry B.; Ekambaram, Rajmadhan; Pratt, Gill A.; Krotkov, Eric; Hackett, Douglas; Yang, Ruigang; Zheng, Qinfen; Sharma, Rajeev; Anderson, Mark; Peot, Mark A.; Aguilar, Mario; Khosla, Deepak; Yang, Chen; Kim, Kyungnam; Elazary, Lior; Voorhies, Randolph; Parks, Daniel; Itti, Laurent

doi:10.1109/icpr.2014.416

Cited by 13 publications

(16 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An image sequence data set that does contain multiple object types has been provided by the DARPA Neovision2 [27] program. This data set was collected to enable training and evaluation of Neuromorphic Vision algorithms [28], [29], [30], [31], which are a class of object recognition algorithms motivated by the emergence of bio-inspired vision sensors [32] and processing hardware (e.g. [33]).…”

Section: Related Workmentioning

confidence: 99%

Track Everything: Limiting Prior Knowledge in Online Multi-Object Recognition

Wong

Stamatescu

Gatt³

et al. 2017

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

This paper addresses the problem of online tracking and classification of multiple objects in an image sequence. Our proposed solution is to first track all objects in the scene without relying on object-specific prior knowledge, which in other systems can take the form of hand-crafted features or user-based track initialization. We then classify the tracked objects with a fast-learning image classifier, that is based on a shallow convolutional neural network architecture and demonstrate that object recognition improves when this is combined with object state information from the tracking algorithm. We argue that by transferring the use of prior knowledge from the detection and tracking stages to the classification stage, we can design a robust, general purpose object recognition system with the ability to detect and track a variety of object types. We describe our biologically inspired implementation, which adaptively learns the shape and motion of tracked objects, and apply it to the Neovision2 Tower benchmark data set, which contains multiple object types. An experimental evaluation demonstrates that our approach is competitive with the state-of-the-art video object recognition systems that do make use of object-specific prior knowledge in detection and tracking, while providing additional practical advantages by virtue of its generality.

show abstract

Section: Related Workmentioning

confidence: 99%

Track Everything: Limiting Prior Knowledge in Online Multi-Object Recognition

Wong

Stamatescu

Gatt³

et al. 2017

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

show abstract

“…This section describes how the new track IDs in Neovision2 Tower may be used to evaluate multi-class, multi-object tracking systems. To date, the Neovision2 detection metrics 8 have been applied to the original Neovision2 Tower data set to evaluate object detection systems by considering all objects as one target class, and object recognition systems, which evaluates the detection of each object class separately.…”

Section: Usage Of the Augmented Ground Truthsmentioning

confidence: 99%

“…In this paper, we describe the development of a data set for evaluating multi-class, multi-object tracking algorithms. This is achieved by augmenting the existing Neovision2 Tower data set, which was created specifically for online object recognition 8 . The resulting data set enables the evaluation of both multi-object tracking and online recognition algorithms.…”

Section: Introductionmentioning

confidence: 99%

A data set for evaluating the performance of multi-class multi-object video tracking

et al. 2017

View full text Add to dashboard Cite

One of the challenges in evaluating multi-object video detection, tracking and classification systems is having publically available data sets with which to compare different systems. However, the measures of performance for tracking and classification are different. Data sets that are suitable for evaluating tracking systems may not be appropriate for classification. Tracking video data sets typically only have ground truth track IDs, while classification video data sets only have ground truth class-label IDs. The former identifies the same object over multiple frames, while the latter identifies the type of object in individual frames. This paper describes an advancement of the ground truth meta-data for the DARPA Neovision2 Tower data set to allow both the evaluation of tracking and classification. The ground truth data sets presented in this paper contain unique object IDs across 5 different classes of object (Car, Bus, Truck, Person, Cyclist) for 24 videos of 871 image frames each. In addition to the object IDs and class labels, the ground truth data also contains the original bounding box coordinates together with new bounding boxes in instances where un-annotated objects were present. The unique IDs are maintained during occlusions between multiple objects or when objects re-enter the field of view. This will provide: a solid foundation for evaluating the performance of multi-object tracking of different types of objects, a straightforward comparison of tracking system performance using the standard Multi Object Tracking (MOT) framework, and classification performance using the Neovision2 metrics. These data have been hosted publically.

show abstract

“…The physiology of the low-level human/mammalian visual systems was characterised by their tests with feline striate cortex as shown in Figure 1 [4]. Several theories and algorithms in image processing, object recognition and computer vision have been developed from this experiment and its findings [5][6][7], [12]. In addition to mimicking the brain, the research on intelligent video signal processing has been widely inspired by neural networks, such as an IR video based early aircraft detection using Kohonen's neural network [6].…”

Section: Neuromorphic Visual Processing Algorithmmentioning

confidence: 99%

Application of neuromorphic visual processing in railway environment

Han

2015

2015 IEEE International Conference on Industrial Technology (ICIT)

View full text Add to dashboard Cite

Considering the number of fatalities and serious injuries, there is a surprising dearth of human figure detection in the context of railway environment. Our proposed neuromorphic visual processing algorithm based on the biological vision system is an effective approach for making detection of human figures in the railway environment based on either stationary image sensor or mobile image sensor. The tests showed that real-life applications could be sought from incorporating neuromorphic visual processing into railway industry for the purpose of safety and security measures, with the reliable performance of cyclist detection from Rail Runner's accident.

show abstract

Performance Evaluation of Neuromorphic-Vision Object Recognition Algorithms

Cited by 13 publications

References 17 publications

Track Everything: Limiting Prior Knowledge in Online Multi-Object Recognition

Track Everything: Limiting Prior Knowledge in Online Multi-Object Recognition

A data set for evaluating the performance of multi-class multi-object video tracking

Application of neuromorphic visual processing in railway environment

Contact Info

Product

Resources

About