Dear All,
I am able to run all the projects on my own, now to explore further : How to do the torch.load of new dataset consist of our “own” png or npy images ? — not the sprites images as in the example given by Professor.
regards
1 Like
| Step | Task |
|---|---|
| 1 | Create or collect sprite images (e.g., 16×16 or 32×32) |
| 2 | Save images as PNGs |
| 3 | Assign labels to each image |
| 4 | Convert to .npy arrays: sprites.npy and labels.npy |
| 5 | Load arrays into the training script and train the model |