Towards Event-Driven Object Detection with Off-the-Shelf Deep Learning

Iacono, Massimiliano; Weber, Stefan; Glover, Arren; Bartolozzi, Chiara

doi:10.1109/iros.2018.8594119

Cited by 42 publications

(33 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They are maybe the type of deep neural networks most widely used in vision applications, see e.g. [13]- [15], [34], [35], and also in event-based perception [22], [36]- [38]. Using CNNs for event classification requires a dense representation of events, which generally are temporally and spatially sparse.…”

Section: Deep Learning Methods For Event-based People Identificationmentioning

confidence: 99%

“…A CNN was trained in [21] using event images and grayscale images to guide a non-holonomic Unmanned Ground Vehicle in a predator-prey experiment. An iCub robot was presented in [22] where event images were used for object detection using an off-the-shelf Deep Learning approach. However, motion and location of aerial vehicles produce different event streams than on ground vehicles.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Event-based human intrusion detection in UAS using Deep Learning

Pérez-Cutiño

Eguíluz

Dios

et al. 2021

2021 International Conference on Unmanned Aircraft Systems (ICUAS)

View full text Add to dashboard Cite

Automatic intrusion detection in unstructured and complex environments using autonomous Unmanned Aerial Systems (UAS) poses perception challenges in which traditional techniques are severely constrained. Event cameras have high temporal resolution and dynamic range, which make them robust against motion blur and lighting conditions. This paper presents an event-by-event processing scheme for detecting human intrusion using UAS. It includes: 1) one method for detecting clusters of events caused by moving objects in static background; and 2) one method based on Convolutional Neural Networks to compute the probability that a cluster corresponds to a person. The proposed scheme has been implemented and validated in challenging scenarios.

show abstract

Section: Deep Learning Methods For Event-based People Identificationmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Event-based human intrusion detection in UAS using Deep Learning

Pérez-Cutiño

Eguíluz

Dios

et al. 2021

2021 International Conference on Unmanned Aircraft Systems (ICUAS)

View full text Add to dashboard Cite

show abstract

“…An end‐to‐end object detector has been proposed for DVS in [13], where the feature extraction network is designed by considering unique data structure from DVS; moreover, the adaptive temporal pooling is applied to balance triggered events between rapid and slow motions. A comparison of object detection performance between traditional image frames and event frames has been presented in [14], which shows the advantages of event‐based approaches in fast and low‐light conditions. Authors in [15] propose two neural network architectures for object detection based on the YOLO [5] detector.…”

Section: Related Workmentioning

confidence: 99%

Object tracking on event cameras with offline–online learning

Jiang¹,

Mou²,

Shi³

et al. 2020

CAAI trans. intell. technol.

View full text Add to dashboard Cite

Compared with conventional image sensors, event cameras have been attracting attention thanks to their potential in environments under fast motion and high dynamic range (HDR). To tackle the lost‐track issue due to fast illumination changes under HDR scene such as tunnels, an object tracking framework has been presented based on event count images from an event camera. The framework contains an offline‐trained detector and an online‐trained tracker which complement each other: The detector benefits from pre‐labelled data during training, but may have false or missing detections; the tracker provides persistent results for each initialised object but may suffer from drifting issues or even failures. Besides, process and measurement equations have been modelled, and a Kalman fusion scheme has been proposed to incorporate measurements from the detector and the tracker. Self‐initialisation and track maintenance in the fusion scheme ensure autonomous real‐time tracking without user intervene. With self‐collected event data in urban driving scenarios, experiments have been conducted to show the performance of the proposed framework and the fusion scheme.

show abstract

“…Since event-based vision is relatively new, only a limited amount of work addresses object detection using these devices (Liu et al, 2016;Iacono et al, 2018;Lenz et al, 2018). Liu et al (2016) focuses on combining a frame-based CNN detector to facilitate the event-based module.…”

Section: Introductionmentioning

confidence: 99%

“…Liu et al (2016) focuses on combining a frame-based CNN detector to facilitate the event-based module. We argue that using intensity images, either reconstructed from the event stream (Scheerlinck et al, 2018) or captured simultaneously (Liu et al, 2016;Iacono et al, 2018), with deep neural networks for event-based object detection may achieve good performance with lots of training data and computing power, but they go against the idea of lowlatency, low-power event-based vision. In contrast, Lenz et al (2018) presents a practical event-based approach to face detection by looking for pairs of blinking eyes.…”

Section: Introductionmentioning

confidence: 99%