YOLOP is an efficient multi-task network for panoptic driving perception that jointly performs car detection, drivable area segmentation, and lane line segmentation with a shared encoder and three task-specific decoders, delivering strong accuracy–efficiency trade-offs for real-time onboard vision.

Original paper: YOLOP: You Only Look Once for Panoptic Driving Perception

YOLOP

This model is the end-to-end YOLOP checkpoint trained for joint detection and segmentation on BDD100K. It is well suited for on-device panoptic driving perception where low latency and power efficiency are critical.

Model Configuration:

• Reference implementation: YOLOP
• Original Weight: End-to-end.pth
• Resolution: 3x384x640
• Support Cooper version:
  • Cooper SDK: [2.5.4]
  • Cooper Foundry: [2.3]

Model	Device	Compression	Model Link
YOLOP	N1-655	Amba_optimized	Model_Link
YOLOP	N1-655	Activation_fp16	Model_Link
YOLOP	X7	Amba_optimized	Model_Link
YOLOP	X7	Activation_fp16	Model_Link
YOLOP	CV7	Amba_optimized	Model_Link
YOLOP	CV7	Activation_fp16	Model_Link
YOLOP	CV72	Amba_optimized	Model_Link
YOLOP	CV72	Activation_fp16	Model_Link
YOLOP	CV75	Amba_optimized	Model_Link
YOLOP	CV75	Activation_fp16	Model_Link