You will get Real-Time Computer Vision Debugging Consultation
Rising Talent
Rising Talent
Project details
Is your computer vision system slow, unstable, or failing on real camera data?
I help diagnose production issues in real-time video AI systems, including RTSP pipelines, OpenCV/C++ or Python bugs, ALPR, face recognition, object detection/tracking, TensorRT/ONNX deployment, NVIDIA Jetson, Docker, and edge AI setups.
This consultation is best if you need a senior technical opinion before committing to a larger implementation.
I can help with:
• RTSP latency, dropped frames, or unstable camera streams
• OpenCV/C++ or Python pipeline bugs
• Models that work offline but fail in production
• Slow inference on GPU, Jetson, or edge devices
• TensorRT, ONNX Runtime, CUDA, or Docker deployment issues
• ALPR, face recognition, detection, or tracking instability
• Camera calibration, stereo vision, or image-processing problems
You get a focused technical call, review of your setup, diagnosis of likely bottlenecks, and clear recommended next steps.
Before the call, please send your system description, current issue, hardware, software stack, and any useful logs, screenshots, code snippets, or sample videos.
I help diagnose production issues in real-time video AI systems, including RTSP pipelines, OpenCV/C++ or Python bugs, ALPR, face recognition, object detection/tracking, TensorRT/ONNX deployment, NVIDIA Jetson, Docker, and edge AI setups.
This consultation is best if you need a senior technical opinion before committing to a larger implementation.
I can help with:
• RTSP latency, dropped frames, or unstable camera streams
• OpenCV/C++ or Python pipeline bugs
• Models that work offline but fail in production
• Slow inference on GPU, Jetson, or edge devices
• TensorRT, ONNX Runtime, CUDA, or Docker deployment issues
• ALPR, face recognition, detection, or tracking instability
• Camera calibration, stereo vision, or image-processing problems
You get a focused technical call, review of your setup, diagnosis of likely bottlenecks, and clear recommended next steps.
Before the call, please send your system description, current issue, hardware, software stack, and any useful logs, screenshots, code snippets, or sample videos.
Machine Learning Tools
OpenCV, Python, PyTorchWhat's included $150
These options are included with the project scope.
$150
- Delivery Time 1 day
- Number of Revisions 0
Optional add-ons
You can add these on the next page.
Written technical report
(+ 1 Day)
+$100
Extra 60-minute call
+$150
Code review before call
+$150
1 review
(1)
(0)
(0)
(0)
(0)
This project doesn't have any reviews.
AM
Artur M.
Apr 19, 2025
C++ Developer for OpenCV and Raspberry Pi Project
Very good and experienced professional, and it was a great pleasure working with.
Highly recommended!!
Highly recommended!!
About Luis Filipe
Computer Vision Consultant | Real-Time Video AI, TensorRT, Edge Deploy
Vicosa, Brazil - 11:32 am local time
Most computer vision projects do not fail because the model cannot be trained.
They fail because the system is too slow, unstable, hard to deploy, or unreliable on real-world camera data.
That’s where I can help.
I work on the full computer vision system:
* RTSP/video ingestion pipelines
* Real-time object detection and tracking
* ALPR and face recognition systems
* Camera calibration and OpenCV pipeline debugging
* Edge AI deployment with NVIDIA Jetson, TensorRT, ONNX Runtime, and Docker
* Model optimization with pruning, quantization, FP16/INT8, and knowledge distillation
* Backend APIs for CV systems using FastAPI, AWS, and vector databases like Milvus
Recent production work:
* Developed a C++ / OpenCV stereo computer vision system to measure real-world interpupillary distance from stereo imagery, including camera calibration, rectification, triangulation, and metric 3D measurement. Achieved ~63.4 mm measured pupil distance, closely matching the ground-truth measurement of 64 mm.
* Deployed a face recognition platform across 40+ locations, handling 400+ concurrent requests with a C++ edge client, Python backend, AWS ECS, and Milvus.
* Built a video-text retrieval platform for natural-language search over weeks of archived footage, reducing model size by 90%, increasing inference speed 4×, scaling camera throughput 6×, and reaching sub-100ms query latency.
* Improved an ALPR pipeline by 20% using per-character matching and majority voting, making results more stable under partial occlusion.
* Built model optimization tooling for pruning and ONNX quantization, improving throughput across 10+ production computer vision models.
I can help if you are dealing with:
* RTSP latency, dropped frames, or unstable video streams
* Models that work offline but fail in production
* Slow inference on edge devices or cloud GPUs
* ALPR, face recognition, tracking, or detection accuracy issues
* OpenCV calibration, stereo vision, or image-processing pipeline problems
* Deployment issues involving Docker, AWS, TensorRT, ONNX, or Jetson devices
For uncertain projects, I usually recommend starting with a short feasibility/debugging audit. I can review your data, code, camera setup, or current pipeline and give you a clear technical diagnosis before you commit to a larger implementation.
Tech stack:
Python, C++, OpenCV, PyTorch, TensorRT, ONNX Runtime, CUDA, Docker, FastAPI, AWS, Milvus, GStreamer, NVIDIA Jetson.
Send me a message with your current setup, dataset, or pipeline issue, and I’ll suggest the fastest path to a working solution.
Steps for completing your project
After purchasing the project, send requirements so Luis Filipe can start the project.
Delivery time starts when Luis Filipe receives requirements from you.
Luis Filipe works on your project following the steps below.
Revisions may occur after the delivery date.
Review your setup
I’ll review your issue description, hardware, software stack, logs, screenshots, code snippets, sample videos, or benchmarks before the call.
Diagnose the issue
During the call, we’ll identify the most likely bottlenecks or failure points in your computer vision, RTSP, OpenCV, TensorRT, or edge AI pipeline.