Datasets drive vision progress, yet existing driving datasets are impoverished in terms of visual content and supported tasks to study multitask learning for autonomous driving. Researchers are usually constrained to study a small set of problems on one dataset, while real-world computer vision applications require performing tasks of various complexities. We construct BDD100K, the largest driving video dataset with 100K videos and 10 tasks to evaluate the exciting progress of image recognition algorithms on autonomous driving. The dataset possesses geographic, environmental, and weather diversity, which is useful for training models that are less likely to be surprised by new conditions. Based on this diverse dataset, we build a benchmark for heterogeneous multitask learning and study how to solve the tasks together. Our experiments show that special training strategies are needed for existing models to perform such heterogeneous tasks. BDD100K opens the door for future studies in this important venue. More detail is at the dataset home page.

Dataset Card for bdd100k-validation

From one of the largest open source driving datasets, BDD100k, is the BDD100K images dataset. The dataset consists of every 10th second in the videos and contains a train, validation and test split. It contains labels for object detection, weather, time of day, and scene of the driving!

This is a FiftyOne dataset with 10000 samples.

  Installation

If you haven't already, install FiftyOne: pip install -U fiftyone

  Usage… See the full description on the dataset page: https://huggingface.co/datasets/dgural/bdd100k.

BDD100K-weather is a dataset which is inherited from BDD100K using image attribute labels for Out-of-Distribution object detection. All images in BDD100K are categorized into six domains, including clear, overcast, foggy, partly cloudy, rainy and snowy. Clear and overcast are used for training while the rest is used for testing, moreover, per training domain is sampled 1.5k images at most while per testing domain is sampled 0.5k images at most. Thus, we have BDD100K-weather (paper is under review).

BDD100K

## Overview

BDD100K is a dataset for object detection tasks - it contains Cars annotations for 5,411 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [Public Domain license](https://creativecommons.org/licenses/Public Domain).

Bdd100k

## Overview

Bdd100k is a dataset for object detection tasks - it contains Vehicles annotations for 9,797 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Dataset

This dataset was created by Nagaraj Madamshetti

Contents

hirundo-io/bdd100k-validation-only dataset hosted on Hugging Face and contributed by the HF Datasets community

BDD100K, A Large-scale Diverse Driving Video Database. in May 2018 Berkeley AI Lab (BAIR) released BDD100K, the largest publicly available driving dataset with the most diverse content, and also designed an image annotation system. the BDD100K dataset contains 100,000 high-definition videos, each video is about 40 seconds \ 720p \ 30 fps. The key frames are sampled at the 10th second of each video to obtain 100,000 images (image size: 1280*720), which are annotated with 10 categories of object bounding boxes, drivable areas, lane markers and full-frame instance segmentation. The dataset is geographically, environmentally and weather diverse. The dataset is divided into 70,000/10,000/20,000 for training/validation/testing respectively. For more information see https://doc.bdd100k.com/index.html.

The images in this package are the frames at the 10th second in the videos. The split of train, validation, and test sets are the same with the whole video set. They are used for object detection, drivable area, lane marking.

This dataset contains only labels for period classification.

hirundo-io/bdd100k-val dataset hosted on Hugging Face and contributed by the HF Datasets community

BDD100k Training

## Overview

BDD100k Training is a dataset for object detection tasks - it contains Person Car Motor Tsign annotations for 10,000 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [MIT license](https://creativecommons.org/licenses/MIT).

BDD100K Object_Detection

## Overview

BDD100K Object_Detection is a dataset for object detection tasks - it contains Objects annotations for 99,750 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Evaluation of GAAN in BDD100K.

Dataset

This dataset was created by Jeremy26

Released under Apache 2.0

Contents

clairexu198/bdd100k dataset hosted on Hugging Face and contributed by the HF Datasets community

In the context of intelligent driving, pedestrian detection faces challenges related to low accuracy in target recognition and positioning. To address this issue, a pedestrian detection algorithm is proposed that integrates a large kernel attention mechanism with the YOLOV5 lightweight model. The algorithm aims to enhance long-term attention and dependence during image processing by fusing the large kernel attention module with the C3 module. Furthermore, it addresses the lack of long-distance relationship information in channel and spatial feature extraction and representation by introducing the Coordinate Attention mechanism. This mechanism effectively extracts local information and focused location details, thereby improving detection accuracy. To improve the positioning accuracy of obscured targets, the alpha CIOU bounding box regression loss function is employed. It helps mitigate the impact of occlusions and enhances the algorithm’s ability to precisely localize pedestrians. To evaluate the effectiveness of trained model, experiments are conducted on the BDD100K pedestrian dataset as well as the Pascal VOC dataset. Experimental results demonstrate that the improved attention fusion YOLOV5 lightweight model achieves an average accuracy of 60.3%. Specifically, the detection accuracy improves by 1.1% compared to the original YOLOV5 algorithm, and the accuracy performance index reaches 73.0%. These findings strongly indicate the proposed algorithm in significantly enhancing the accuracy of pedestrian detection in road scenes.

Dataset

This dataset was created by Ahmed Stohy Muhammed

Contents

The BDD100k dataset is a large-scale driving dataset containing 100,000 images of various weather conditions, including clear, rainy, foggy, and night-time.

Bdd100k Bus

## Overview

Bdd100k Bus is a dataset for object detection tasks - it contains Bus annotations for 10,243 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Traffic object detection results: Comparing on the KITTI dataset.

epiphanicc/bdd100k dataset hosted on Hugging Face and contributed by the HF Datasets community