Machine Learning Baby Monitor Prevents The Hunger Games

Newborn babies can be tricky to figure out, especially for first-time parents. Despite the abundance of unsolicited advice proffered by anyone who ever had a baby before — and many who haven’t — most new parents quickly get in sync with the baby’s often ambiguous signals. But [Caleb] took his observations of his newborn a step further and built a machine-learning hungry baby early warning system that’s pretty slick.

Normally, babies are pretty unsubtle about being hungry, and loudly announce their needs to the world. But it turns out that crying is a lagging indicator of hunger, and that there are a host of face, head, and hand cues that precede the wailing. [Caleb] based his system on Google’s MediaPipe library, using his baby monitor’s camera to track such behaviors as lip smacking, pacifier rejection, fist mouthing, and rooting, all signs that someone’s tummy needs filling. By putting together a system to recognize these cues and assign a weight to them, [Caleb] now gets a text before the baby gets to the screaming phase, to the benefit of not only the little nipper but to his sleep-deprived servants as well. The video below has some priceless bits in it; don’t miss [Baby Caleb] at 5:11 or the hilarious automatic feeder gag at the end.

We’ve seen some interesting videos from [Caleb] recently, mostly having to do with his dog’s bathroom habits and getting help cleaning up afterward. We can only guess how those projects will be leveraged when this kid gets a little older and starts potty training.

25 thoughts on “Machine Learning Baby Monitor Prevents The Hunger Games

  1. Caleb here, nice write up! This was a fun project & works really well. Hah, gotta start thinking about how unnecessary amounts of technology can help w/ potty training as he grows up.

    Also, getting ahead of the more observant parents out there: I shouldn’t have used the clip of my baby in loose blankets. I recognize this is unsafe, and he does not sleep/nap for any extended period of time like that. I was in the same room as him when that shot was captured

  2. Wow. ML can actually be useful? I’m shocked. Watching all demos that say “hey there is a banana” were so boring. I know there is a banana, so what; “well I can count how many bananas are there”; good for you if there ever was a job about counting bananas it would be obsolete.

    This is so nice and practical. I can maybe persuade my kids to produce offsprings.

    Thanks Caleb, really nicely done.

  3. Careful about getting too clever with ML. You might wind up with a toddler who gets along great with the machine but not necessarily the parents.

    Crying is part of how an infant learns to communicate with you and vice versa. It’s unpleasant, but don’t be in too great a hurry to suppress it.

    1. This. My boy would sound the alarm when it was time to eat and his face would light up when we came in the room, that’s how we mainly knew he was hungry-lonely. Fantastic (and finally useful) application of machine learning though

      1. Yeah who knows what all the downstream implications could be. I’m not claiming it’s all positive. But imo if it can enable me to feed our baby once at night w/o wifey waking up, it’s a win. Don’t think it would be wise to rely on it too much beyond that

        1. (Facetious Mode: On)
          Yeah, you don’t want the child to grow up expecting things done or given to him without asking!
          B^)
          (Facetious Mode: Off)

    2. I have no doubt the child will learn to use crying for communication anyway. Still plenty of reasons to denote displeasure. Crying is however a bad method of communication and even very young children (starting at roughly 6 months) can very quickly learn non-verbal (sign) language to communicate all sorts of things (look up baby sign language). Teaching, encouraging and understanding non-verbal communications is much more beneficial than “crying means you want/get attention”.

  4. Feed a little and feed often, direct from a breast, as natural selection has engineered babies to work. My wife and I have had to deal with 5 infants, including two sets of twins with one set being premature, so we are very experience. It is a huge sacrifice of your time and energy, no tech can get around that.

  5. This is a great idea I love it help stop baby become very stress when hungry and allowance mother to get much needed rest after giving birth. Then can take trun to let each parents get good night sleep, with out stressing when the baby will wake up hungry. Well done this is amazing I would definitely invest in a camera like this if I was have another baby.

  6. New born routines are like clockworks, eat, burp, sleep. No need for AI what new parents need is stamina to care for the newborn every 3 to 4 hours that it need food.

  7. I posted this and your robotic laser guided poop patrol posts to my humor list. Interestingly enough it started a conversation where one of my friends suggested that you apply the same techniques to recognizing precursors of Sudden Infant Death Syndrome (SIDS). That developing a SIDS monitor and early warning system could literally be a live saver.

    1. Or how to get sued 6 ways from Sunday, for the deaths it didn’t prevent, and the “false alarms” that “needlessly” keep waking up some parents with particularly SIDsy babies…. maybe it saved their life 23 times, but no, gotta be a faulty product. :-/

      1. Last I heard SIDS is not completely understood (just did an online search and some group in AU last month claimed to have found a potential cause), and there are also ways to work around most of the 6+ ways you mention. The sad fact is that anyone can sue ANYONE for anything they want to for anything they think they can get away with. Are their warning signs that are suggestive of a potential issue? Is there liabilities for false positives? Are their liabilities for false negatives? But my comment ans questions are for Caleb — have you thought about searching/training for non-safe sleep conditions and other things? I bet for his personal use he would love to know when something is going on which is dangerous. There is also the issue of people following on his work, or using his code, is he legally liable if he is looking into things? That is for him to decide RW. What he chooses to work on next.

        1. Yeah I’m not trying to get sued for a weekend project that I thought would be useful occasionally at night to help my wife sleep. Existing expensive AI backed monitors already do a bit of safety checks, e.g. room temp, etc.. I wouldn’t doubt that there are already things that check for a covered mouth

        2. Yeah, but it’s a particular tough area to do anything in, because you are now entering… the emotive zone… there’s a “local pulse” group that I’m on that has phases of going full on mommy group, and they’ll get real offended on there if you suggest co sleeping is bad for instance, “It’s my kid of course I’m not going to kill him, I love him, are you accusing me of not loving my kid??? Reeeeeee!!!! “

Leave a Reply to EnricoCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.