Using Yolov5 + CRNN， half-finshed

dataset

orginal data:label.csv ./data
data for yolo: ./datayolo ./yolodata
data for CRNN: ./dataCRNN ./CRNN/lmdb*-->data turned into lmdb
- using ./utils to generate the dataset for the two

train

according to official github

weight trained

yolo:./weights/yolov5_best.pth
CRNN:./CRNN/expr2/best.pth

intergrated demo.py for implement

./demo.py

input path: ./data1/
ouput csv: ./result.csv

Usage

Usage:

python generate_labels.py -d <path> -s <path> -f <format>

Options:

-d, --data_path Specifies the path where the data is stored.

-s, --save_path Specifies the path where the output will be saved.

-f, --format Specifies the file format to process, supported formats: {bench, aig}.
Note: Files in .aiger format must be converted to .aig format before use.

Conversion command example：
```
 for file in *.aiger; do mv -- "$file" "${file%.aiger}.aig"; done
```

Example

python generate_labels.py -d ~/AIGDataset/all_case/aig -s ~/AIGDataset/all_case -f aig

Explanation: This command reads all files in ~/AIGDataset/all_case/aig（including all corresponding files in subdirectories, e.g., ~/AIGDataset/all_case/aig[/a/b/xxx.aig]）and saves the results in corresponding subdirectories, e.g., ~/AIGDataset/all_case[/a/b/xxx/counting_simplify_PO/]. Meanwhile, an abc.log file will be created in the specified save_path to record the output logs of abc.

In the counting_simplify_PO directory, the following files will be generated:

.
├── signed_edge.csv
├── node-type.csv
└── label.csv

File Description

signed_edge.csv: Each row represents <start_id>, <end_id>, {1,-1}, where 1 indicates an edge without a not, and -1 indicates an edge with a not.
node-type.csv: The rows are ordered by node IDs. Each row indicates the type of the node, where 0 represents a primary input (PI), and 1 represents an AND gate.
label.csv: The first row indicates <number of FA>, <number of HA> (FA: Full Adder, HA: Half Adder). Subsequent rows, ordered by node IDs, indicate whether the corresponding node is <FAin>, <FAout>, <HAin>, <HAout>, represented as boolean values.

Hinataee/CoursesML