Build pipelines that merge deep learning with conventional computer vision methods for the following applications:- Crowd panic detection- Human fall detection- License plate recognition- Smoke detection- Vehicle detectionMain skills:- Object detection, segmentation, and classification- Object tracking- Keypoints estimation- Crowd estimation- Optical flow estimation

Embedded Computer Vision.- Research the state of the art of object detection models, including the lightweight models.- Train, evaluate and integrate the first version of Neuroo smoke detection model into the Axis camera.- Test and measure the performance of several modern object detection models on the Axis camera.• Tools: C/C++, Axis ACAP, Python, Tensorflow Object Detection API, TFLite, Pytorch, Docker.

De-identification of personal data in multi-media content.- Research and develop the state of the art of whole body de-identification based on GAN and VAE.- Implement clothing style and license plate de-identification algorithms.- Develop first version of the XXII Pseudonymisation Toolkit capable of pseudonymizing faces, clothing styles, license plates, specific zones.- Edit instruction document.• Tools: Python, Tensorflow/Keras, Pytorch, Docker, Celery/Redis.