Car Detection Model Zoo

Documentation Models
1

Dataset

Dataset Name Training Images Validation Images Class Labels License
Open Images Dataset 20,000 [1] 5,095 0: Car Annotations under CC BY 4.0, Images under CC BY 2.0

Model Zoo

DeGirum’s car detection model zoo offers models trained with the Ultralytics repository.

Model Architecture Input Size Precision Runtime Device Type mAP 50-95 mAP 50 FPS
yolov5nu_relu6_car 640x640 INT8 N2X ORCA1 0.6637 0.8458 116.5
yolov5nu_relu6_car 640x640 INT8 TFLITE EDGETPU 0.6640 0.8455 –
yolov5nu_relu6_car 640x640 INT8 OPENVINO CPU 0.6686 0.8476 –
yolov5nu_relu6_car 640x640 FP32 OPENVINO CPU 0.6721 0.8505 –
yolov5nu_silu_car 640x640 INT8 N2X ORCA1 0.6590 0.8394 –
yolov5nu_silu_car 640x640 INT8 TFLITE EDGETPU 0.6584 0.8393 –
yolov5nu_silu_car 640x640 INT8 OPENVINO CPU 0.6695 0.8452 –
yolov5nu_silu_car 640x640 FP32 OPENVINO CPU 0.6745 0.8486 –
yolov8n_relu6_car 640x640 INT8 N2X ORCA1 0.6784 0.8513 107.7
yolov8n_relu6_car 640x640 INT8 TFLITE EDGETPU 0.6789 0.8517 –
yolov8n_relu6_car 640x640 INT8 OPENVINO CPU 0.6833 0.8551 –
yolov8n_relu6_car 640x640 FP32 OPENVINO CPU 0.6853 0.8562 –
yolov8n_silu_car 640x640 INT8 N2X ORCA1 0.6723 0.8484 –
yolov8n_silu_car 640x640 INT8 TFLITE EDGETPU 0.6725 0.8483 –
yolov8n_silu_car 640x640 INT8 OPENVINO CPU 0.6827 0.8547 –
yolov8n_silu_car 640x640 FP32 OPENVINO CPU 0.6886 0.8574 –
yolov5su_relu6_car 640x640 INT8 N2X ORCA1 0.6846 0.8592 43.5
yolov5su_relu6_car 640x640 INT8 TFLITE EDGETPU 0.6844 0.8581 –
yolov5su_relu6_car 640x640 INT8 OPENVINO CPU 0.6948 0.8643 –
yolov5su_relu6_car 640x640 FP32 OPENVINO CPU 0.6967 0.8642 –
yolov5su_silu_car 640x640 INT8 N2X ORCA1 0.6850 0.8578 –
yolov5su_silu_car 640x640 INT8 TFLITE EDGETPU 0.6849 0.8567 –
yolov5su_silu_car 640x640 INT8 OPENVINO CPU 0.6962 0.8645 –
yolov5su_silu_car 640x640 FP32 OPENVINO CPU 0.7000 0.8650 –
yolov8s_relu6_car 640x640 INT8 N2X ORCA1 0.6916 0.8600 38.0
yolov8s_relu6_car 640x640 INT8 TFLITE EDGETPU 0.6919 0.8600 –
yolov8s_relu6_car 640x640 INT8 OPENVINO CPU 0.6966 0.8633 –
yolov8s_relu6_car 640x640 FP32 OPENVINO CPU 0.7025 0.8675 –
yolov8s_silu_car 640x640 INT8 N2X ORCA1 0.6894 0.8603 –
yolov8s_silu_car 640x640 INT8 TFLITE EDGETPU 0.6893 0.8602 –
yolov8s_silu_car 640x640 INT8 OPENVINO CPU 0.7005 0.8678 –
yolov8s_silu_car 640x640 FP32 OPENVINO CPU 0.7041 0.8698 –

  1. A subset of car training images are used for training. ↩︎