Browse categories
Explore
Fiverr Pro
English
$
USD
I will build custom computer vision, object detection, and face rcognition models
About this Gig
Looking for a Computer Vision Engineer to develop custom object detection models, face recognition systems, or real-time video analytics? You are in the right place!
I specialize in building end-to-end computer vision solutions using PyTorch, OpenCV, YOLOv8, and vector search indexing. Whether you need custom keypoint detection, multi-camera CCTV monitoring, or high-accuracy image classification, I deliver optimized, production-grade solutions.
What I Offer:
- Object Detection & Keypoint Tracking: Custom YOLOv8/PyTorch model training for objects, keypoints, and precise visual transcription.
- Face Recognition & Surveillance: Sub-100ms identity matching pipelines using Hugging Face embeddings (FaceNet) + FAISS vector search.
- Real-Time Video Analytics: Multi-camera CCTV stream processing (25+ FPS), frame-drop optimization, and automated alert engines (SMS/Email).
- Hybrid Deep Learning Models: Vision Transformers (Swin Transformer), CNNs, feature extraction, and zero-shot anomaly detection.
- Full-Stack CV Integration: Flask web applications, Docker containerization, and cloud deployment.
Why Choose Me?
- Proven Field Experience: Engineered AI surveillance systems handling concurrent CCT
APIs:
Other
Programming language:
Python
•
SQL
•
Colab
•
NoSQL
Tools:
Jupyter Notebook
•
OpenCV
•
TensorFlow
•
MLflow
•
CVAT
•
Colab
•
PyTorch
Frameworks:
Scikit-learn
•
Keras
•
PyTorch
•
Panda
Other Data Science & ML Services I Offer
FAQ
Can you train YOLOv8 on my custom dataset?
Yes! I can annotate or process your dataset, train a custom YOLOv8 model for object or keypoint detection, and evaluate performance (mAP, recall, precision).
How fast is the face recognition / identity matching system?
By integrating deep feature embeddings with FAISS vector indexing, identity matching occurs in under 100ms across 500+ registered profiles without lagging video streams.
Can your system handle live CCTV or multi-camera feeds?
Yes. I build real-time video processing pipelines in OpenCV that process multi-stream video at 25+ FPS without frame drops.
Will I receive the complete model weights and source code?
Absolutely. You will get full Python source code, model weight files (.pt, .onnx, or .h5), setup instructions, and clean documentation.
