Computer Vision Engineer Needed for Object Detection and Image Processing System

We're looking for an experienced Computer Vision Engineer to develop an advanced object detection and image processing system capable of real-time analysis, recognition, and classification of visual data.

Project Overview:

Build a computer vision solution for automated quality inspection and defect detection in manufacturing. The system should process video streams in real-time, detect anomalies, classify defects, and provide instant alerts with high accuracy.

Key Responsibilities:

Develop object detection and recognition algorithms

Implement image preprocessing and enhancement techniques

Train and fine-tune deep learning models for visual tasks

Build real-time video processing pipelines

Implement multi-object tracking systems

Create image segmentation and classification models

Optimize models for edge deployment (NVIDIA Jetson, Raspberry Pi)

Develop calibration and camera integration solutions

Build annotation tools and data pipelines

Create REST APIs for model inference

Implement quality metrics and performance monitoring

Deploy models to production environments

Required Skills:

3+ years of computer vision and deep learning experience

Expert proficiency in OpenCV and image processing techniques

Strong knowledge of deep learning frameworks (TensorFlow, PyTorch)

Experience with object detection models (YOLO, Faster R-CNN, SSD, EfficientDet)

Image segmentation experience (U-Net, Mask R-CNN, DeepLab)

Python programming and numerical computing (NumPy, SciPy)

Understanding of CNN architectures and transfer learning

Experience with model optimization (TensorRT, ONNX, OpenVINO)

Camera calibration and 3D reconstruction knowledge

Edge device deployment experience

Technical Stack:

Languages: Python 3.8+, C++ (optional)

Computer Vision: OpenCV 4.x, PIL/Pillow

Deep Learning: TensorFlow 2.x, PyTorch, Keras

Object Detection: YOLO v5/v8, Detectron2, MMDetection

Model Optimization: TensorRT, ONNX Runtime, TFLite

Image Processing: scikit-image, albumentations

Annotation Tools: LabelImg, CVAT, Labelbox

Deployment: Docker, FastAPI, Flask

Hardware: NVIDIA Jetson (Nano/Xavier), Intel NCS, Raspberry Pi

Computer Vision Tasks:

Real-time object detection and tracking

Image classification and recognition

Semantic/instance segmentation

Defect detection and quality inspection

Optical character recognition (OCR)

Face detection and recognition (if applicable)

Pose estimation and gesture recognition

Image enhancement and restoration

Anomaly detection

Model Requirements:

High accuracy (>95% for critical detections)

Real-time inference (<50ms per frame)

Robust to varying lighting conditions

Handle occlusions and multiple objects

Scalable to different camera resolutions

Low false positive rate

Edge-optimized for resource constraints

Data & Training:

Dataset collection and annotation strategy

Data augmentation techniques

Class imbalance handling

Train/validation/test split methodology

Transfer learning from pre-trained models

Custom dataset creation and labeling

Active learning implementation

Deployment Environment:

Edge devices (NVIDIA Jetson, Raspberry Pi)

Cloud deployment (AWS, GCP, Azure)

RTSP/IP camera integration

Multi-camera synchronization

GPU acceleration (CUDA)

REST API for inference

Real-time streaming capabilities

Performance Optimization:

Model quantization (INT8, FP16)

Pruning and compression techniques

Batch processing optimization

Multi-threading for video processing

GPU memory management

Inference speed optimization

Deliverables:

Trained computer vision models with high accuracy

Complete source code with documentation

Model training pipeline and scripts

Data preprocessing and augmentation code

Real-time inference API (REST/gRPC)

Edge deployment package (Docker/binaries)

Performance benchmarking report

Dataset annotation guidelines

Model evaluation metrics and confusion matrix

Deployment and integration guide

User documentation and API reference

Budget: $55 - $110/hour (Hourly) or $10,000 - $22,000 (Fixed project)

Timeline: 8-14 weeks