For Safer Navigation: Pedestrian-View Intersection Classification

Astrid, Marcella; Lee, Jin Ha; Zaheer, Muhammad Zaigham; Lee, Jae-Yeong; Lee, Seung-Ik

doi:10.1109/ictc49870.2020.9289182

Cited by 4 publications

(10 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We collected our data in urban areas using phone and action cameras. In theory, the problem addressed by this paper can be adequately addressed with pedestrian‐view data, as discussed in Astrid et al [1]. Moreover, because there are no public pedestrian‐view intersection classification datasets, our proposed technique can be beneficial in reducing the cost of data collection and labeling.…”

Section: Methodsmentioning

confidence: 98%

“…Notably, when the agent enters an intersection, it must take appropriate actions, such as turning or continuing forward. Several studies have provided methods of classifying intersections using still images or videos [1][2][3][4][5][6]; most of these studies use one-camera data. However, the one-camera approach fails to detect the intersection when the agent is inside the intersection as it can no longer detect the notable features, as illustrated in Figure 1.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Assembling three one‐camera images for three‐camera intersection classification

Astrid,

Lee

2023

ETRI Journal

Self Cite

View full text Add to dashboard Cite

Determining whether an autonomous self‐driving agent is in the middle of an intersection can be extremely difficult when relying on visual input taken from a single camera. In such a problem setting, a wider range of views is essential, which drives us to use three cameras positioned in the front, left, and right of an agent for better intersection recognition. However, collecting adequate training data with three cameras poses several practical difficulties; hence, we propose using data collected from one camera to train a three‐camera model, which would enable us to more easily compile a variety of training data to endow our model with improved generalizability. In this work, we provide three separate fusion methods (feature, early, and late) of combining the information from three cameras. Extensive pedestrian‐view intersection classification experiments show that our feature fusion model provides an area under the curve and F1‐score of 82.00 and 46.48, respectively, which considerably outperforms contemporary three‐ and one‐camera models.

show abstract

Section: Methodsmentioning

confidence: 98%

mentioning

confidence: 99%

Assembling three one‐camera images for three‐camera intersection classification

Astrid,

Lee

2023

ETRI Journal

Self Cite

View full text Add to dashboard Cite

show abstract

“…The task is to classify whether a selected the image was captured at an intersection. The most straightforward method would be a data-driven approach wherein a large number of intersection and non-intersection labeled images are prepared to train deep neural networks, as performed in [2].…”

Section: Proposed Methodsmentioning

confidence: 99%

“…However, even a human would hesitate to label this location as an intersection because of the absence of clearly sectioned road. Conventional intersection detection studies [2,3,4,19] did not encounter this ambiguity as they only focused on on-board cameras or pedestrian To address this problem, we redefined the intersection itself more rigorously and defined the problem accordingly. Specifically, an intersection in a 360 • image was defined as a location where there are multiple PDoTs.…”

Section: Proposed Methodsmentioning

confidence: 99%

“…In addition, as this model accepts multiple video frames as input, its robustness in dynamic scenes with several walking people, such as in urban shopping malls, is questionable. On the other hand, Astrid et al [2] recently proposed a single-image intersection classification system using ResNetbased architecture trained on a pedestrian-view-level image dataset containing 345 and 498 intersection and non-intersection images, respectively. Although the system achieved a test accuracy of 80%, the training and test data comprised pedestrian-view images of sidewalks with little or no pedestrian traffic, and the generalization of this method to disparate test scenes from the training dataset is still not confirmed.…”

Section: Traffic Intersection Detectionmentioning

confidence: 99%

See 1 more Smart Citation

Intersection Prediction from Single 360° Image via Deep Detection of Possible Direction of Travel

Sugimoto¹,

Ikehata²,

Aizawa³

2022

Preprint

View full text Add to dashboard Cite

Movie-Map, an interactive first-person-view map that engages the user in a simulated walking experience, comprises short 360 • video segments separated by traffic intersections that are seamlessly connected according to the viewer's direction of travel. However, in wide urban-scale areas with numerous intersecting roads, manual intersection segmentation requires significant human effort. Therefore, automatic identification of intersections from 360 • videos is an important problem for scaling up Movie-Map. In this paper, we propose a novel method that identifies an intersection from individual frames in 360 • videos. Instead of formulating the intersection identification as a standard binary classification task with a 360 • image as input, we identify an intersection based on the number of the possible directions of travel (PDoT) in perspective images projected in eight directions from a single 360 • image detected by the neural network for handling various types of intersections. We constructed a large-scale 360 • Image Intersection Identification (iii360) dataset for training and evaluation where 360 • videos were collected from various areas such as school campus, downtown, suburb, and china town and demonstrate that our PDoT-based method achieves 88% accuracy, which is significantly better than that achieved by the direct binary classification based method. The source codes and a partial dataset will be shared in the community after the paper is published.

show abstract