3 ms·
Show HN: Caloric Lab - AI Photo Calorie Counter
- hartator 10y agoI've coded it, ask me anything. We have used several machine learning models trained on different kind of food. We have used Caffe (a machine learning framework) to train on images. First identify the food, then we try to evaluate the features of the food. Like greasiness. The challenges were that iPhones have limited amount of RAM available and we have to modify the way Caffe works to make everything fit in the memory.
- fiftyacorn 10y agoGreat work - how does it measure against photos from cookery websites where they tell you the calories?
- hartator 10y agoThe app technical architecture is still pretty minimal. I’ve been running only 2 models. One for identifying the food. One for giving a scale if the food is greasy or lean. Coupled to average calorie database, a small algorithm is then able to do basic calorie estimation. The model to identify food can identify between 97 different foods. The greasiness/leanness model is still pretty raw. It has only been trained only on around 500 images.