Jigsaw Jeeves
A Python library that acts as a computer-vision assistant for solving jigsaw puzzles. Rather than solving the puzzle for you, it tells you where each piece most likely belongs, narrowing the search space to a small ranked candidate list so you can make progress when you are stuck.
Installation
pip install jigsaw-jeeves
Requires Python 3.12 or above.
Dependencies (opencv-python, numpy, scipy) are installed automatically.
Prerequisites
To use Jigsaw Jeeves you need two images:
- Reference image -- a photo of the puzzle box cover (the completed picture)
- Scrambled image -- a photo of the puzzle pieces arranged face-up in a neat rectangular grid on a flat, contrasting surface
Both images should be reasonably well-lit and photographed from roughly overhead. The grid dimensions you pass to solve() must match how you physically arranged the pieces before photographing.
Quick Start
from jigsaw_jeeves import solve
results = solve(
reference_image_filepath="box_cover.jpg",
scrambled_image_filepath="pieces_on_table.jpg",
grid=(20, 25), # rows x cols matching how you arranged the pieces
top_k=3, # number of candidate positions to return per piece
save_to_file="solution.txt", # optional: also write results to a file
)
# results maps each scrambled tile position to a ranked list of candidates.
for pos, candidates in results.items():
best_dest, score = candidates[0]
print(f"Piece at {pos} most likely belongs at {best_dest} (similarity: {score:.3f})")
solve() -- full signature
solve(
reference_image_filepath: str,
scrambled_image_filepath: str,
grid: tuple[int, int] | None = None,
top_k: int = 3,
save_to_file: str | None = None,
) -> dict
Parameters
| Parameter | Type | Description |
|---|---|---|
reference_image_filepath |
str |
Path to the puzzle box cover image (the solved reference). |
scrambled_image_filepath |
str |
Path to a photograph of the scrambled pieces arranged face-up in a rectangle on a flat surface. |
grid |
tuple[int, int] or None |
(rows, cols) matching the physical layout of pieces. If None, inferred automatically from the reference image dimensions (targeting tiles of ~150×150 px). |
top_k |
int |
Number of candidate destination positions to return per tile (default 3). |
save_to_file |
str or None |
If provided, results are written to this file path. The parent directory must exist. The dict is always returned regardless. |
Returns
A dict mapping each scrambled tile position (row, col) to a ranked list of up to top_k candidate solved positions. Each list entry is ((dest_row, dest_col), cosine_similarity_score). The first entry is the globally optimal assignment (via the Hungarian algorithm); remaining entries are the next-best matches by cosine similarity.
Raises
ValueError with a human-friendly message if:
- Either image path does not exist or cannot be opened
gridis not a 2-tuple of positive integerstop_kis not a positive integersave_to_file's parent directory does not exist- Either image is entirely black after background suppression (insufficient contrast with background)
How It Works
- The pipeline overlays an R-by-C grid on both images and treats each grid cell as the unit of comparison.
- Each cell is represented as a 513-dimensional feature vector: a normalized 3-D RGB color histogram (512 values) and a single edge-density scalar.
- Cosine similarity is used to measure how closely a scrambled tile matches each reference tile.
- The Hungarian algorithm finds the globally optimal bijective assignment, guaranteeing that every scrambled tile is matched to a unique reference position.
License
MIT
Metadata
Release files for jigsaw-jeeves 0.0.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| jigsaw_jeeves-0.0.2.tar.gz | 9.7 kB | Without provenance | Details for jigsaw_jeeves-0.0.2.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| jigsaw_jeeves-0.0.2-py3-none-any.whl | Python 3 | none | any | Details for jigsaw_jeeves-0.0.2-py3-none-any.whl |
Total release size: 20.8 kB
Release files / jigsaw_jeeves-0.0.2.tar.gz
| Download URL | jigsaw_jeeves-0.0.2.tar.gz |
|---|---|
| Size | 9.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
de93150a19dfbc90da9226f978c16f767505314688028f84c844b4982ec0fc7d
|
|
BLAKE2b-256 checksum How to use checksums |
30a951a7d7ac6d11e7dd47c1e93da14fc5584df3200063c68912d1899a16536e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.8.11
|
Release files / jigsaw_jeeves-0.0.2-py3-none-any.whl
| Download URL | jigsaw_jeeves-0.0.2-py3-none-any.whl |
|---|---|
| Size | 11.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
ee604cc9e7263813fd66ec6972695df3cd1f35b007c2b00e8a20861d4af81421
|
|
BLAKE2b-256 checksum How to use checksums |
7b81947ea8e1dff522dffb751f49bb5b6626d3eda437825715edf8e5b7578a44
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.8.11
|
Security for jigsaw-jeeves 0.0.2
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.