Automated shopping experience using cashier-less systems

A method of associating items with initial measurements and stock keeping unit (SKU) numbers, comprising:

• receiving a SKU number for an item and iteration number on a device, wherein the SKU number identifies a particular item, wherein the device comprises at least one sensor;
• determining a first unit measurement of the item when the particular item corresponding to the received SKU number is placed on the at least one sensor from a particular bin of a shelving unit;
• determining a second unit measurement of the item when the particular item from the at least one sensor is picked up from the at least one sensor from the particular bin of the shelving unit;
• and
• determining subsequent unit measurement when the particular item corresponding to the received SKU number is placed on the at least one sensor and when the particular item corresponding to the received SKU number is picked up from the at least one sensor for a number of times, wherein the number of times corresponds to the received iteration number.

A method of configuring a device with a sensor using machine learning, comprising:

• obtain a data stream from the sensor on the device;
• detecting a measurement change from the sensor, wherein the measurement change exceeds a pre-defined value;
• capturing a key signal frame within a predefined time interval based on determining that the measurement change exceeds the pre-defined value, wherein the key signal frame comprises a measurement signal having a first measurement reading, the detected measurement change, and a second measurement reading after the detected measurement change;
• generating metadata comprising the captured measurement signal;
• and
• storing the captured measurement signal and generated metadata on a pre-trained model on the device.

A method of identifying an item using a sensor and machine learning, comprising:

• obtaining a data stream from the sensor coupled to a device;
• detecting a measurement change from the sensor, wherein the measurement change exceeds a pre-defined value;
• and
• based on determining that the measurement change exceeds the pre-defined value,
• capturing a key signal frame within a predefined time interval, wherein the key signal frame comprises a measurement signal having a first measurement reading, the detected measurement change, and a second measurement reading after the detected measurement change,
• clustering the detected measurement change into a plurality of clusters from a pre-trained clustering model,
• determining whether the detected measurement change matches a measurement value from a cluster among the plurality of clusters from the pre-trained clustering model, and
• outputting results from the clustering based on a determination that the detected measurement change matches the measurement value from the cluster among the plurality of clusters from the pre-trained clustering model.