Open-Vocabulary Detection with OWLv2 on Jetson Orin
Full OWLv2 inference pipeline on Jetson Orin with Triton Inference Server for prompt-based open-vocabulary detection.
Duration
2024 - Now
Client
AWENTIA
Tech stack
Computer VisionOWLv2JetsonTriton
- Implemented the full OWLv2 inference pipeline on Jetson Orin using Triton Inference Server for edge deployment.
- Enabled prompt-based open-vocabulary detection from text or image queries.
- Optimized serving and post-processing to achieve 4 FPS in production-like conditions.