Vision integration of 3D fruit and vegetable scanner with reinforcement model

To feed our intelligent algorithms with useful data, we need large amounts of stackable products. To cope with the large varieties in fruit and vegetables, we need to find out how we can convert 3D vision information of these products to high-quality lowkey input information to use as input in the stacking algorithms and training data. In this internship you will research how we can reliably acquire this information with 3D vision techniques, convert it to useful information for the training of our stacking algorithm and test this method.

Job description

• Perform research on methods for adding low-key data to existing reinforcement learning networks.

• Develop a method to convert 3D vision data to low-key data to use as stack-algorithm input.

• Develop the training model and build a test environment for this.

• Combine and implement your model within the main software and perform tests.

• Optimize the performance based on test results.

What are we looking for?

• You are currently doing a Bachelor or Master in engineering, computers sciences or equivalent experience.

• You have experience in developing machine vision applications/setups

• You have experience with one or more AI/vision libraries (e.g. Keras, PyTorch, OpenCV, Halcon).

• You are not afraid to dive into research literature.

• You are available for 4+ months.

• Experience with software development (C#.NET, C++, Python) is a big plus.

• Fluency in English, Dutch is a plus.