I cannot access your link, I guess you should ask Microsoft or Kaggle how you can download it.
Then to import that in Edge Impulse studio, make sure your dataset is not too big otherwise you’ll be limited using the free licence in Edge Impulse Studio.
I’d suggest you try to import something like 100 to 200 data samples, train a first model, and increase your dataset size and train again until you reach your needed accuracy or the max job time limit.
yes, I did download ’ Microsoft or Kaggle’, but there are only ‘cats and dogs’, where can I find other pictures like any things?
You can have a look at the COCO dataset: COCO - Common Objects in Context
However, for real production use cases, I’d suggest you try to collect and annotate your dataset yourself. I know it is a hard work but it will suits exactly your need.
Feel free to share more insights about your use case here too.
I didn’t figure out how to download yet.
BTW. I’m using Arduino IDE, there are lot of errors came out when compiling sketch relative to Edge impulse, do I need to install any thing? or what can be the reason?
fatal error: base/cef_build.h: No such file or directory
error: #error "Atomic operations are not supported on your platform"
do I involved in cross compiling? can I avoid this?
Hello @SHADAM ,
Can I make sure you’ve selected the right board in Arduino IDE?