Internship in computer vision development, detection and direction estimation.

Company Description

Convinced that technology must contribute to making our environments simpler, more sustainable and safer, LACROIX (5300 employees, €619M in proforma turnover) supports its clients in the construction and management of intelligent living ecosystems, thanks to connected equipment and technologies and through a vision:

"Connected Technologies for a Smarter World".

At the heart of the L25 group strategy, Impulse is the R&D Business Unit of our Electronics activity. It brings together key technologies to support companies and industrialists in speeding up their technological and digital transformation. The Impulse offering relies on a complete centre of expertise in the design and industrialisation of IoT solutions (hardware, software and cloud), AI, Computer Vision, Cybersecurity, ... to become a leading player in Industrial IoT. Impulse targets markets with high societal stakes and growth prospects, especially in the fields of home automation, industry, mobility and automotive.

Mainly based in Cesson in the heart of the Rennes technology basin, the Impulse teams (currently 100 employees) are destined to grow with LACROIX by contributing to its growth.

Job Description

Subject: Object Detection and Estimation of Direction and Velocity from Static POV

Attached to Ian, Software Engineer, within the R&D team at the heart of our design office, the purpose of the internship is twofold.

1. In general, study multi-task learning, through a relatively simple application of vehicle detection on two consecutive frames, with an estimation of their direction of movement and speed. It will be necessary to study how to keep the majority of the algorithm/network common to the two tasks.
2. Examine the state of the art which will serve as inspiration for techniques and layers. The design of the algorithm will be done from scratch starting with the simplest architecture, and increasing complexity gradually to achieve the expected performances. The last constraint will be to make an estimation of runtime, and to stay below a target defined by the manager of the tutor.

Your missions:

• Study of the state of the art and feasibility
  • Neural Networks, "classic" Computer Vision or a combination of the two
• Planning of metrics and expected performances
• Study to maximize the mutualization of operations and information
• Estimation of runtime on proprietary accelerators with maximum runtime limit (looping if necessary)
• Creation of the project from scratch and training
• Evaluation and validation of results
• Optional:
  • Same missions from a moving vehicle
  • Addition of the estimation of the direction and speed of the camera as 4th and 5th outputs
• Expected results:
  • Trained network
  • Up-to-date GIT with rigor and commented
  • Performance estimation report (estimated runtime and accuracy)
  • Example results

Qualifications

• Student in engineering school or equivalent, you are said to be a proactive person with a spirit of creativity.
• You have a real passion for algorithms, especially in computer vision.
• You have good skills in Python programming and deep learning and knowledge of embedded C would be a plus.
• Experience in a framework is required (Caffe, ONNX, Pytorch, Keras...)
• You have a fluent level of English.
• Minimum availability of 4 months but the internship is better suited for a duration of 6 months.

Additional Information

Benefits:

• Paid internship with access to restaurant vouchers
• Opportunity to work on innovative and high-tech projects
• Supervision by an experienced and passionate team