Repo for the final project of DIO Microsoft Copilot AI bootcamp
The images in the "inputs" directory where feed one by one to Microsoft Copilot via web browser, with the following prompt: "Tell me what's written on the image"
The results where copied and pasted on the directory "outputs", each one in the correspondent file.
- Big letters with contrasting colors are easily read by Copilot
- Copilot ignores some figures as texts, as the 3D "+" on image2 text. Where we read "Treine +", Copilot only reads "Treine"
- It not only shows the text on the image, but tends to understand it's context, even trying to make a conversation out of it
- Copilot also recognizes objects on the image and mentions them, even when not being explicitly asked to.
- It may overlook smallsized and secondary texts, like the timestamp on image3.jpeg
- It includes emojis on the answer 😁