Machine Learning PoC for counting "items" on conveyor

We need a willing person to label items in provided images (approximately 100) to then train these on small MobileNet v4 (or similar) models for counting purposes. Stretch goal, to synchronise detections between systems, identify a reoccurring detection. 100% payout would be on 26 repeatable tests with a success would be a 97% or higher on production counting tasks. 50% payout on several demonstrations of the equipment. Things to know: - supplied live camera feed or test video stream will be available - labelling images will be supplied - example labels and how to train the model will be provided - research on how to do the task and/or improvements would be expected If it's found to achieve success was never going to happen with our assumptions, effort payment on the final 50% is up to the discretion of the team. Meaning if you put in the effort in and demonstrate that it is somewhat possible to achieve given more resources, we'll probably pay out the other 50%