Vehicle Exterior Pet Recognition Image Dataset

Description: This dataset contains high-quality synthetic images of cats and dogs captured within a close range of 0.5–1.5 meters from the exterior of a vehicle, using fisheye cameras configured for sentinel mode perception. It covers a variety of pet appearances, including light-colored and dark-colored individuals of different sizes, under both daytime and nighttime lighting conditions. The images are generated to reflect natural animal positions and movements in real-world vehicle surroundings, with a resolution of 768×512 pixels.

The dataset is designed for animal recognition in sentinel mode systems, supporting applications such as vehicle perimeter monitoring, small animal detection for collision prevention, and near-range object classification. By simulating realistic approach scenarios of cats and dogs around parked vehicles, it enables robust model training for accurate species and appearance classification.

All samples are annotated with object detection bounding boxes in JSON format, covering pet categories such as cat (light/dark) and dog (light/dark), with environmental tags for daytime and nighttime scenarios. The dataset ensures balanced representation across lighting conditions, pet color variations, and body sizes, making it suitable for training models that generalize well across diverse urban and suburban environments.

The synthetic generation process incorporates multiple ground surface types, background textures, and lighting variations, along with natural animal postures, to simulate a wide range of real-world situations. Capture angles include front, side, and rear perspectives relative to the vehicle, enhancing detection robustness in sentinel vision pipelines.

By combining photorealistic rendering, detailed human annotation, and diverse scenario simulation, this dataset provides a strong foundation for developing high-accuracy pet detection models, improving vehicle safety alerts, and enhancing intelligent surveillance capabilities.

Keywords: pet recognition sentinel mode cat dog detection near-range perception animal classification

Sector: Intelligent Transportation / Surveillance / Animal Detection Systems / Computer Vision

Metrics: Pet detection accuracy, species classification precision Entities: Light-colored cats, dark-colored cats, light-colored dogs, dark-colored dogs Geographies: China (urban and residential environments) Price Range (USD): 4000–20000 Source type: Synthetic data generation from simulation environment, human annotation Sample Images: