Spaces:
Sleeping
Sleeping
File size: 95,101 Bytes
afa662f 0ee82c6 afa662f 0ee82c6 afa662f 53311be afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 1b20296 0ee82c6 16216eb 0ee82c6 16216eb 0ee82c6 16216eb 0ee82c6 16216eb 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 0ee82c6 afa662f 11b7b0a afa662f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 |
import streamlit as st
import numpy as np
import cv2
from PIL import Image
from io import BytesIO
from ultralytics import YOLO
#import ollama
from datetime import datetime
import tempfile
import os
import base64
import bcrypt
import sqlite3
import time
from kokoro import KPipeline
import soundfile as sf
from IPython.display import Audio
import torch
from googletrans import Translator
from sentence_transformers import SentenceTransformer
#from ragas.metrics import AnswerRelevancy, Faithfulness, AnswerCorrectness, ContextPrecision
#from ragas import evaluate
#from ragas.metrics import (
# answer_relevancy,
# faithfulness,
# answer_correctness,
# context_precision
#)
from datasets import Dataset
import pandas as pd
import random
import plotly.express as px
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from typing import List, Dict
import asyncio
from ragas.embeddings.base import BaseRagasEmbeddings
from dotenv import load_dotenv
from groq import Groq
# Load environment variables from .env file
load_dotenv()
# Initialize Groq client
groq_client = Groq(api_key=os.environ['GROQ_API_KEY'])
class SentenceTransformerEmbeddings(BaseRagasEmbeddings):
"""
A wrapper class to adapt SentenceTransformer to the BaseRagasEmbeddings interface.
This class implements both synchronous and asynchronous embedding methods.
"""
def __init__(self, model_name: str = "sentence-transformers/all-MiniLM-L6-v2"):
self.model = SentenceTransformer(model_name)
def embed_query(self, text: str) -> List[float]:
"""
Embed a single query (text) using SentenceTransformer (synchronous).
"""
return self.model.encode(text).tolist()
def embed_documents(self, texts: List[str]) -> List[List[float]]:
"""
Embed a list of documents (texts) using SentenceTransformer (synchronous).
"""
return [self.model.encode(text).tolist() for text in texts]
async def aembed_query(self, text: str) -> List[float]:
"""
Embed a single query (text) using SentenceTransformer (asynchronous).
"""
loop = asyncio.get_event_loop()
return await loop.run_in_executor(None, self.embed_query, text)
async def aembed_documents(self, texts: List[str]) -> List[List[float]]:
"""
Embed a list of documents (texts) using SentenceTransformer (asynchronous).
"""
loop = asyncio.get_event_loop()
return await loop.run_in_executor(None, self.embed_documents, texts)
class RAGSystemVariants:
def __init__(self):
self.embedding_model = SentenceTransformer('all-MiniLM-L6-v2')
async def baseline_rag(self, query, top_k=3):
"""Your current full RAG system"""
chunks = retrieve_relevant_documents(query, [], top_k)
context = "\n".join([chunk["text"] for chunk in chunks])
response = await generate_groq_response(f"Context: {context}\n\nQuestion: {query}")
return response, context
async def no_retrieval(self, query):
"""Generation only - no retrieval"""
response = await generate_groq_response(query)
return response, ""
async def random_retrieval(self, query, top_k=3):
"""Random document selection instead of semantic retrieval"""
try:
all_docs = client.scroll(collection_name=collection_name, limit=100)[0]
if len(all_docs) > 0:
random_chunks = random.sample(all_docs, min(top_k, len(all_docs)))
context = "\n".join([chunk.payload["text"] for chunk in random_chunks])
else:
context = ""
response = await generate_groq_response(f"Context: {context}\n\nQuestion: {query}")
return response, context
except Exception as e:
st.error(f"Error in random retrieval: {e}")
return "Error in random retrieval", ""
async def different_top_k(self, query, top_k):
"""Test different top-k values"""
chunks = retrieve_relevant_documents(query, [], top_k)
context = "\n".join([chunk["text"] for chunk in chunks])
response = await generate_groq_response(f"Context: {context}\n\nQuestion: {query}")
return response, context
def create_test_dataset(limit=20):
"""Create a test dataset for RAGAS evaluation"""
test_cases = []
try:
conn = sqlite3.connect('./db/disease_knowledge_base.db')
c = conn.cursor()
c.execute("SELECT name, cause, symptoms, treatment FROM diseases LIMIT ?", (limit,))
diseases = c.fetchall()
conn.close()
for disease_name, cause, symptoms, treatment in diseases:
questions_and_answers = [
(f"What causes {disease_name}?", cause),
(f"What are the symptoms of {disease_name}?", symptoms),
(f"How do I treat {disease_name}?", treatment),
(f"Tell me about {disease_name}", f"Cause: {cause}\nSymptoms: {symptoms}\nTreatment: {treatment}"),
]
for question, ground_truth in questions_and_answers:
test_cases.append({
"question": question,
"ground_truth": ground_truth,
"disease": disease_name
})
return test_cases[:limit]
except Exception as e:
st.error(f"Error creating test dataset: {e}")
return []
async def run_ablation_study(progress_bar, status_text, max_questions=20):
"""Run comprehensive ablation study with progress tracking"""
status_text.text("Creating test dataset...")
test_cases = create_test_dataset(limit=max_questions)
if not test_cases:
st.error("No test cases created. Check your database connection.")
return None
rag_variants = RAGSystemVariants()
experiments = {
"Full_RAG_k3": lambda q: rag_variants.baseline_rag(q, top_k=3),
"No_Retrieval": lambda q: rag_variants.no_retrieval(q),
"Random_Retrieval": lambda q: rag_variants.random_retrieval(q, top_k=3),
"RAG_k1": lambda q: rag_variants.different_top_k(q, top_k=1),
"RAG_k5": lambda q: rag_variants.different_top_k(q, top_k=5),
"RAG_k10": lambda q: rag_variants.different_top_k(q, top_k=10),
}
all_results = []
total_experiments = len(experiments) * len(test_cases)
current_progress = 0
for exp_name, exp_func in experiments.items():
status_text.text(f"Running experiment: {exp_name}")
questions = []
answers = []
contexts = []
ground_truths = []
for test_case in test_cases:
try:
answer, context = await exp_func(test_case["question"])
questions.append(test_case["question"])
answers.append(answer)
contexts.append([context] if context else [""])
ground_truths.append(test_case["ground_truth"])
current_progress += 1
progress_bar.progress(current_progress / total_experiments)
except Exception as e:
st.error(f"Error in {exp_name}: {e}")
continue
exp_results = []
evaluator = LocalMetricsEvaluator()
for q, a, c, gt in zip(questions, answers, contexts, ground_truths):
context_str = c[0] if c and c[0] else ""
metrics = {
"question": q,
"answer": a,
"context": context_str,
"ground_truth": gt,
"experiment": exp_name,
"answer_relevancy": evaluator.evaluate_answer_relevancy(q, a),
"faithfulness": evaluator.evaluate_faithfulness(a, context_str) if context_str else 1.0,
"answer_correctness": evaluator.evaluate_answer_correctness(a, gt),
"context_precision": evaluator.evaluate_context_precision(q, context_str) if context_str else 0.0,
"context_recall": evaluator.evaluate_context_recall(q, context_str, gt) if context_str else 0.0
}
exp_results.append(metrics)
all_results.extend(exp_results)
return pd.DataFrame(all_results)
def visualize_ablation_results(results_df):
"""Create interactive visualizations for ablation study results"""
summary_stats = results_df.groupby('experiment').agg({
'answer_relevancy': ['mean', 'std'],
'faithfulness': ['mean', 'std'],
'answer_correctness': ['mean', 'std'],
'context_precision': ['mean', 'std'],
'context_recall': ['mean', 'std']
}).round(3)
summary_stats.columns = ['_'.join(col).strip() for col in summary_stats.columns.values]
summary_stats = summary_stats.reset_index()
metrics = ['answer_relevancy_mean', 'faithfulness_mean', 'answer_correctness_mean',
'context_precision_mean', 'context_recall_mean']
# Radar chart
fig_radar = go.Figure()
for _, row in summary_stats.iterrows():
fig_radar.add_trace(go.Scatterpolar(
r=[row[metric] for metric in metrics],
theta=[metric.replace('_mean', '').replace('_', ' ').title() for metric in metrics],
fill='toself',
name=row['experiment']
))
fig_radar.update_layout(
polar=dict(
radialaxis=dict(
visible=True,
range=[0, 1]
)),
showlegend=True,
title="RAGAS Metrics Comparison Across Experiments"
)
# Bar chart comparison
fig_bar = make_subplots(
rows=2, cols=3,
subplot_titles=[metric.replace('_mean', '').replace('_', ' ').title() for metric in metrics],
)
for i, metric in enumerate(metrics):
row = (i // 3) + 1
col = (i % 3) + 1
fig_bar.add_trace(
go.Bar(
x=summary_stats['experiment'],
y=summary_stats[metric],
error_y=dict(type='data', array=summary_stats[metric.replace('mean', 'std')]),
name=metric.replace('_mean', '').replace('_', ' ').title(),
showlegend=False
),
row=row, col=col
)
fig_bar.update_layout(height=800, title="Detailed Metrics Comparison")
return fig_radar, fig_bar, summary_stats
def render_research_page():
"""Render the research/ablation study page"""
st.title("🔬 RAG System Research Dashboard")
st.markdown("Systematic evaluation and ablation study of the crop disease detection RAG system")
# Initialize session state for results
if 'ablation_results' not in st.session_state:
st.session_state['ablation_results'] = None
tabs = st.tabs(["Ablation Study", "Model Comparison", "Error Analysis", "Export Results"])
with tabs[0]:
st.header("🧪 Ablation Study")
st.write("This systematically evaluates different components of the RAG system.")
col1, col2 = st.columns(2)
with col1:
max_questions = st.number_input("Number of test questions per experiment",
min_value=5, max_value=50, value=20)
with col2:
selected_model_research = st.selectbox(
"Select Model for Experiments",
list(SUPPORTED_MODELS.keys()),
key="research_model_select"
)
if st.button("🚀 Start Ablation Study", type="primary"):
progress_bar = st.progress(0)
status_text = st.empty()
with st.spinner("Running ablation study..."):
try:
results_df = asyncio.run(run_ablation_study(progress_bar, status_text, max_questions))
if results_df is not None:
st.session_state['ablation_results'] = results_df
st.success("✅ Ablation study completed!")
# Show summary statistics
st.subheader("📊 Summary Statistics")
summary_stats = results_df.groupby('experiment').agg({
'answer_relevancy': 'mean',
'faithfulness': 'mean',
'answer_correctness': 'mean',
'context_precision': 'mean',
'context_recall': 'mean'
}).round(3)
st.dataframe(summary_stats, use_container_width=True)
# Key insights
best_overall = summary_stats.mean(axis=1).idxmax()
st.success(f"🏆 **Best Overall Configuration:** {best_overall}")
col1, col2, col3 = st.columns(3)
with col1:
best_relevancy = summary_stats['answer_relevancy'].idxmax()
st.metric("Best Answer Relevancy", best_relevancy,
f"{summary_stats.loc[best_relevancy, 'answer_relevancy']:.3f}")
with col2:
best_faithfulness = summary_stats['faithfulness'].idxmax()
st.metric("Best Faithfulness", best_faithfulness,
f"{summary_stats.loc[best_faithfulness, 'faithfulness']:.3f}")
with col3:
best_correctness = summary_stats['answer_correctness'].idxmax()
st.metric("Best Correctness", best_correctness,
f"{summary_stats.loc[best_correctness, 'answer_correctness']:.3f}")
# Create and display visualizations
fig_radar, fig_bar, summary_stats_detailed = visualize_ablation_results(results_df)
st.subheader("📈 Results Visualization")
viz_tab1, viz_tab2, viz_tab3 = st.tabs(["Radar Chart", "Detailed Comparison", "Raw Data"])
with viz_tab1:
st.plotly_chart(fig_radar, use_container_width=True)
st.markdown("**Interpretation:** The radar chart shows the relative performance of each experiment across all RAGAS metrics. Larger areas indicate better overall performance.")
with viz_tab2:
st.plotly_chart(fig_bar, use_container_width=True)
st.markdown("**Interpretation:** The bar charts show detailed performance with error bars indicating standard deviation across test cases.")
with viz_tab3:
st.dataframe(results_df, use_container_width=True)
# Download options
csv = results_df.to_csv(index=False)
st.download_button(
label="📥 Download Raw Results (CSV)",
data=csv,
file_name=f"ablation_study_results_{datetime.now().strftime('%Y%m%d_%H%M%S')}.csv",
mime="text/csv"
)
except Exception as e:
st.error(f"❌ Error running ablation study: {str(e)}")
st.exception(e)
with tabs[1]:
st.header("Model Comparison")
st.write("Compare different LLM models on the same test dataset.")
selected_models = st.multiselect(
"Select models to compare",
list(SUPPORTED_MODELS.keys()),
default=list(SUPPORTED_MODELS.keys())[:2]
)
num_questions_comp = st.number_input("Number of questions for comparison",
min_value=5, max_value=30, value=10)
if selected_models and st.button("🔍 Run Model Comparison"):
st.info("Model comparison functionality can be extended here...")
progress_bar_comp = st.progress(0)
status_text_comp = st.empty()
with st.spinner("Comparing models..."):
# Create a simplified comparison focusing on generation quality
test_cases = create_test_dataset(limit=num_questions_comp)
comparison_results = []
total_comparisons = len(selected_models) * len(test_cases)
current_progress_comp = 0
for model_name in selected_models:
status_text_comp.text(f"Testing model: {model_name}")
for test_case in test_cases:
try:
# Generate response with current model
response = asyncio.run(generate_groq_response(
test_case["question"],
model_name=SUPPORTED_MODELS[model_name]["name"]
))
# Evaluate
evaluator = LocalMetricsEvaluator()
comparison_results.append({
"model": model_name,
"question": test_case["question"],
"answer": response,
"ground_truth": test_case["ground_truth"],
"disease": test_case["disease"],
"answer_relevancy": evaluator.evaluate_answer_relevancy(test_case["question"], response),
"answer_correctness": evaluator.evaluate_answer_correctness(response, test_case["ground_truth"])
})
current_progress_comp += 1
progress_bar_comp.progress(current_progress_comp / total_comparisons)
except Exception as e:
st.error(f"Error testing {model_name}: {e}")
continue
if comparison_results:
comp_df = pd.DataFrame(comparison_results)
# Summary by model
model_summary = comp_df.groupby('model').agg({
'answer_relevancy': 'mean',
'answer_correctness': 'mean'
}).round(3)
st.subheader("📊 Model Performance Summary")
st.dataframe(model_summary, use_container_width=True)
# Visualization
fig_model_comp = px.bar(
model_summary.reset_index(),
x='model',
y=['answer_relevancy', 'answer_correctness'],
title="Model Performance Comparison",
barmode='group'
)
st.plotly_chart(fig_model_comp, use_container_width=True)
# Store results
st.session_state['model_comparison_results'] = comp_df
with tabs[2]:
st.header("Error Analysis")
st.write("Analyze failure cases and performance patterns.")
if st.session_state['ablation_results'] is not None:
results_df = st.session_state['ablation_results']
# Find worst performing cases
st.subheader("Worst Performing Cases")
worst_cases = results_df.nsmallest(10, 'answer_correctness')[['question', 'answer', 'ground_truth', 'experiment', 'answer_correctness']]
st.dataframe(worst_cases, use_container_width=True)
# Performance by experiment
st.subheader("Performance Distribution")
fig_box = px.box(results_df, x='experiment', y='answer_correctness',
title="Answer Correctness Distribution by Experiment")
st.plotly_chart(fig_box, use_container_width=True)
else:
st.info("Run an ablation study first to see error analysis.")
with tabs[3]:
st.header("Export Results")
st.write("Export results for research papers and further analysis.")
if st.session_state['ablation_results'] is not None:
results_df = st.session_state['ablation_results']
# Generate summary report
report = f"""
# RAG System Ablation Study Report
**Date:** {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}
**Total Experiments:** {len(results_df['experiment'].unique())}
**Total Test Cases:** {len(results_df)}
## Summary Statistics
{results_df.groupby('experiment').agg({
'answer_relevancy': ['mean', 'std'],
'faithfulness': ['mean', 'std'],
'answer_correctness': ['mean', 'std'],
'context_precision': ['mean', 'std'],
'context_recall': ['mean', 'std']
}).round(3).to_string()}
## Best Performing Configurations
- **Best Answer Relevancy:** {results_df.groupby('experiment')['answer_relevancy'].mean().idxmax()}
- **Best Faithfulness:** {results_df.groupby('experiment')['faithfulness'].mean().idxmax()}
- **Best Answer Correctness:** {results_df.groupby('experiment')['answer_correctness'].mean().idxmax()}
## Recommendations
Based on the ablation study results, we recommend...
[Add your analysis here]
"""
col1, col2 = st.columns(2)
with col1:
st.download_button(
label="📄 Download Report (Markdown)",
data=report,
file_name=f"ablation_report_{datetime.now().strftime('%Y%m%d_%H%M%S')}.md",
mime="text/markdown"
)
with col2:
csv_data = results_df.to_csv(index=False)
st.download_button(
label="📊 Download Data (CSV)",
data=csv_data,
file_name=f"ablation_data_{datetime.now().strftime('%Y%m%d_%H%M%S')}.csv",
mime="text/csv"
)
else:
st.info("No results available for export. Run an ablation study first.")
# Database setup
conn = sqlite3.connect('users.db')
c = conn.cursor()
c.execute('''CREATE TABLE IF NOT EXISTS users
(id INTEGER PRIMARY KEY AUTOINCREMENT,
username TEXT UNIQUE,
password_hash TEXT)''')
conn.commit()
# Password hashing and verification
def hash_password(password):
return bcrypt.hashpw(password.encode(), bcrypt.gensalt())
def verify_password(password, hashed_password):
return bcrypt.checkpw(password.encode(), hashed_password)
# Add a user
def add_user(username, password):
# Check if username already exists
c.execute("SELECT id FROM users WHERE username = ?", (username,))
result = c.fetchone()
if result:
return False # Username already exists
# Hash the password and insert the new user
password_hash = hash_password(password)
c.execute("INSERT INTO users (username, password_hash) VALUES (?, ?)",
(username, password_hash))
conn.commit()
return True
# Verify a user
def verify_user(username, password):
c.execute("SELECT password_hash FROM users WHERE username = ?", (username,))
result = c.fetchone()
if result:
return verify_password(password, result[0])
return False
# Login and logout
def login(username, password):
if not username or not password:
st.error("Username and password are required.")
return False
if verify_user(username, password):
st.session_state['authenticated'] = True
st.session_state['username'] = username
st.session_state['last_activity'] = time.time()
return True
st.error("Invalid username or password.")
return False
def logout():
st.session_state['authenticated'] = False
st.session_state['username'] = None
# Add this at the top of your file
def local_css():
st.markdown("""
<style>
.stButton>button {
width: 100%;
border-radius: 5px;
height: 3em;
margin-top: 10px;
}
.auth-container {
max-width: 400px;
margin: auto;
padding: 20px;
border-radius: 10px;
box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
background-color: white;
}
.auth-title {
text-align: center;
font-size: 24px;
margin-bottom: 20px;
color: #1f1f1f;
}
.auth-subtitle {
text-align: center;
font-size: 16px;
margin-bottom: 20px;
color: #666;
}
.hero-section {
text-align: center;
padding: 40px 20px;
background: linear-gradient(to right, #4f46e5, #3b82f6);
color: white;
margin-bottom: 30px;
}
.feature-container {
max-width: 1200px;
margin: auto;
padding: 20px;
display: grid;
grid-template-columns: repeat(auto-fit, minmax(300px, 1fr));
gap: 20px;
margin-bottom: 40px;
}
.feature-card {
background: white;
padding: 20px;
border-radius: 10px;
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
transition: transform 0.3s ease, box-shadow 0.3s ease;
}
.feature-card:hover {
transform: scale(1.05);
box-shadow: 0 4px 8px rgba(0, 0, 0, 0.2);
}
.feature-title {
color: #1f1f1f;
font-size: 18px;
margin-bottom: 10px;
font-weight: bold;
}
.feature-text {
color: #666;
font-size: 14px;
}
.divider {
text-align: center;
margin: 20px 0;
position: relative;
}
.divider:before {
content: "";
position: absolute;
top: 50%;
left: 0;
right: 0;
height: 1px;
background-color: #e0e0e0;
z-index: -1;
}
.divider span {
background-color: white;
padding: 0 10px;
color: #666;
font-size: 14px;
}
@keyframes typing {
0% {
width: 0;
}
50% {
width: 100%;
}
60% {
width: 100%;
}
100% {
width: 0;
}
}
@keyframes blink {
50% {
border-color: transparent;
}
}
.hero-title{
display: inline-block;
font-size: 2.5em;
white-space: nowrap;
overflow: hidden;
border-right: 2px solid white;
width: 0;
animation: typing 6s steps(40, end) infinite, blink 0.5s step-end infinite;
}
.hero-section {
text-align: center;
padding: 40px 20px;
background: linear-gradient(45deg, #4f46e5, #3b82f6);
background-size: 300% 300%;
animation: gradientShift 8s ease infinite;
color: white;
margin-bottom: 30px;
opacity: 0;
animation: fadeIn 2s ease-in-out forwards;
}
@keyframes fadeIn {
from {
opacity: 0;
}
to {
opacity: 1;
}
}
@keyframes gradientShift {
0% {
background-position: 0% 50%;
}
50% {
background-position: 100% 50%;
}
100% {
background-position: 0% 50%;
}
}
/*.feature-container {
display: flex;
justify-content: center;
align-items: center;
gap: 20px;
position: relative;
width: 100%;
height: 300px;
animation: rotate 20s linear infinite; /* Rotate the container */
}
.feature-card {
background: white;
padding: 20px;
border-radius: 10px;
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
transition: transform 0.3s ease, box-shadow 0.3s ease;
flex-shrink: 0;
width: 250px;
}
.feature-card:hover {
transform: scale(1.1);
box-shadow: 0 4px 8px rgba(0, 0, 0, 0.2);
}
@keyframes rotate {
from {
transform: rotate(0deg);
}
to {
transform: rotate(-360deg);
}
*/}
/*.feature-container {
display: flex;
justify-content: center;
align-items: center;
overflow: hidden;
position: relative;
width: 100%;
height: 300px;
}
.feature-track {
display: flex;
animation: circularMove 15s linear infinite;
}
.feature-card {
flex: 0 0 300px; /* Fixed width for each card */
margin: 0 20px;
background: white;
color: #333; /* Text color */
padding: 20px;
border-radius: 10px;
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
text-align: center; /* Center-align the text */
overflow: hidden; /* Prevent overflow issues */
}
.feature-card h3 {
font-size: 1.2em;
margin-bottom: 10px;
text-align: center;
}
.feature-card p {
font-size: 0.9em;
line-height: 1.4;
text-align: center;
font-weight: bold;
}
.feature-card:hover {
transform: scale(1.1);
box-shadow: 0 4px 8px rgba(0, 0, 0, 0.2);
}
@keyframes circularMove {
0% {
transform: translateX(0);
}
100% {
transform: translateX(-100%);
}
*/}
.feature-container {
display: flex;
justify-content: center;
align-items: center;
height: 400px;
perspective: 1000px;
perspective-origin: 50% 50%;
background: linear-gradient(to bottom, #1e293b, #0f172a); /* Dark blue gradient background */
overflow: hidden;
position: relative;
padding: 40px 0;
}
.feature-track {
position: relative;
width: 100%;
height: 100%;
display: flex;
transform-style: preserve-3d;
animation: carousel 15s linear infinite;
}
.feature-card {
position: absolute;
width: 300px;
padding: 50px;
background: white;
border-radius: 15px;
box-shadow: 0 4px 20px rgba(0, 0, 0, 0.3); /* Enhanced shadow for better contrast */
backface-visibility: hidden;
transform-origin: center center;
transition: all 0.5s ease;
}
.feature-card h3 {
color: #1e293b;
font-size: 1.5em;
margin-bottom: 1rem;
font-weight: bold;
}
.feature-card p {
color: #475569;
line-height: 1.6;
}
/* Position and animate cards */
.feature-card:nth-child(1) {
transform: rotateY(0deg) translateZ(400px) translateX(0px);
}
.feature-card:nth-child(2) {
transform: rotateY(60deg) translateZ(400px) translateX(0px);
}
.feature-card:nth-child(3) {
transform: rotateY(120deg) translateZ(400px) translateX(0px);
}
.feature-card:nth-child(4) {
transform: rotateY(180deg) translateZ(400px) translateX(0px);
}
.feature-card:nth-child(5) {
transform: rotateY(240deg) translateZ(400px) translateX(0px);
}
.feature-card:nth-child(6) {
transform: rotateY(300deg) translateZ(400px) translateX(0px);
}
@keyframes carousel {
0% {
transform: translateZ(-400px) rotateY(0deg);
}
100% {
transform: translateZ(-400px) rotateY(-360deg);
}
}
/* Enhanced hover effect with glow */
.feature-card:hover {
transform: scale(1.1) translateZ(450px);
box-shadow: 0 8px 30px rgba(255, 255, 255, 0.1); /* Glowing effect */
z-index: 1;
}
/* Gradient overlays for depth effect */
.feature-container::before,
.feature-container::after {
content: '';
position: absolute;
width: 100%;
height: 100px;
z-index: 2;
pointer-events: none;
}
.feature-container::before {
top: 0;
background: linear-gradient(to bottom, #1e293b, rgba(30, 41, 59, 0));
}
.feature-container::after {
bottom: 0;
background: linear-gradient(to top, #1e293b, rgba(30, 41, 59, 0));
</style>
""", unsafe_allow_html=True)
# Check session expiry
if 'authenticated' in st.session_state and st.session_state['authenticated']:
if time.time() - st.session_state['last_activity'] > 1800: # 30 minutes
logout()
st.rerun()
st.session_state['last_activity'] = time.time()
# Initialize session state for registration form visibility
if 'show_register_form' not in st.session_state:
st.session_state['show_register_form'] = False
# Replace your login/registration section with this:
if 'authenticated' not in st.session_state or not st.session_state['authenticated']:
local_css()
# Landing page hero section
st.markdown("""
<div class="hero-section">
<h1 class="hero-title" style="font-size: 2.5em; margin-bottom: 20px;">Crop Disease Detection System</h1>
<p style="font-size: 1.2em; max-width: 800px; margin: 0 auto;">
An advanced AI-powered system that helps farmers and agricultural experts identify and manage crop diseases effectively
</p>
</div>
""", unsafe_allow_html=True)
# Features section using Streamlit columns
st.subheader("Key Features")
col1, col2, col3 = st.columns(3)
st.markdown("""
<div class="feature-container">
<div class="feature-track">
<div class="feature-card">
<h3>🔍 Instant Detection</h3>
<p>Upload images of your crops and get immediate disease detection results using state-of-the-art AI technology.</p>
</div>
<div class="feature-card">
<h3>💡 Expert Analysis</h3>
<p>Receive detailed analysis and recommendations from our plant pathology expert system.</p>
</div>
<div class="feature-card">
<h3>📊 Detailed Reports</h3>
<p>Generate comprehensive reports with treatment recommendations and preventive measures.</p>
</div>
<div class="feature-card">
<h3>🔍 Instant Detection</h3>
<p>Upload images of your crops and get immediate disease detection results using state-of-the-art AI technology.</p>
</div>
<div class="feature-card">
<h3>💡 Expert Analysis</h3>
<p>Receive detailed analysis and recommendations from our plant pathology expert system.</p>
</div>
<div class="feature-card">
<h3>📊 Detailed Reports</h3>
<p>Generate comprehensive reports with treatment recommendations and preventive measures.</p>
</div>
</div>
</div>
""", unsafe_allow_html=True)
# Crop carousel section
st.markdown("""
<div class="crop-carousel-container">
<div class="crop-carousel-track">
<div class="crop-card">
<img src="https://github.com/ROBERT-ADDO-ASANTE-DARKO/AI-powered-crop-disease-detection/blob/main/images/b034333ddcc732299d45abf753f3fa71f6ff48ffa3338bfecd615bc2.jpg?raw=true" alt="Crop 1">
<h4>Corn Leaf Blight</h4>
<p>Corn leaf blight is a fungal disease caused primarily by Exserohilum turcicum (Northern corn leaf blight) and Bipolaris maydis (Southern corn leaf blight).</p>
</div>
<div class="crop-card">
<img src="https://github.com/ROBERT-ADDO-ASANTE-DARKO/AI-powered-crop-disease-detection/blob/main/images/apple.jpg?raw=true" alt="Crop 2">
<h4>Apple Scab Leaf</h4>
<p>Apple scab is a fungal disease caused by Venturia inaequalis. It primarily affects apple and crabapple trees.</p>
</div>
<div class="crop-card">
<img src="https://github.com/ROBERT-ADDO-ASANTE-DARKO/AI-powered-crop-disease-detection/blob/main/images/tomato.jpg?raw=true" alt="Crop 3">
<h4>Tomato Leaf Late Blight</h4>
<p>Late blight of tomato is caused by the oomycete pathogen Phytophthora infestans. It is characterized by dark, water-soaked lesions on leaves, stems, and fruit.</p>
</div>
<div class="crop-card">
<img src="https://github.com/ROBERT-ADDO-ASANTE-DARKO/AI-powered-crop-disease-detection/blob/main/images/918d1d7a3dda5ce8fbdabf92e5bf38f104efd129ee09adcc6d1ad46c.jpg?raw=true" alt="Crop 4">
<h4>Tomato Leaf Yellow Virus</h4>
<p>Tomato leaf yellow virus (often referred to as Tomato yellow leaf curl virus, or TYLCV) is a viral disease transmitted by whiteflies. It causes yellowing and curling of tomato leaves.</p>
</div>
</div>
</div>
""", unsafe_allow_html=True)
st.markdown("""
<style>
.crop-carousel-container {
width: 100%;
max-width: 800px;
margin: auto;
overflow: hidden;
position: relative;
}
.crop-carousel-track {
display: flex;
animation: moveLeft 20s linear infinite; /* Move right to left */
}
.crop-card {
flex: 0 0 300px;
margin: 0 20px;
background: white;
color: #333;
padding: 20px;
border-radius: 10px;
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
text-align: center;
overflow: hidden;
}
.crop-card img {
width: 100%;
height: 150px;
object-fit: cover;
border-radius: 10px;
margin-bottom: 10px;
}
.crop-card h4 {
font-size: 1.2em;
margin: 10px 0;
}
.crop-card p {
font-size: 0.9em;
line-height: 1.4;
color: #555;
}
@keyframes moveLeft {
0% {
transform: translateX(100%);
}
100% {
transform: translateX(-100%);
}
}
</style>
""", unsafe_allow_html=True)
# Add some spacing
st.markdown("<br>", unsafe_allow_html=True)
# Authentication container
st.markdown('<div class="auth-container">', unsafe_allow_html=True)
# Initialize password reset state
if 'show_reset_form' not in st.session_state:
st.session_state['show_reset_form'] = False
# Update password function
def update_password(username, new_password):
conn = sqlite3.connect('users.db')
c = conn.cursor()
# Check if username exists
c.execute("SELECT id FROM users WHERE username = ?", (username,))
if not c.fetchone():
return False
# Update password
password_hash = bcrypt.hashpw(new_password.encode(), bcrypt.gensalt())
c.execute("UPDATE users SET password_hash = ? WHERE username = ?",
(password_hash, username))
conn.commit()
conn.close()
return True
# Update the authentication container section
if not st.session_state.get('authenticated', False):
st.markdown('<div class="auth-container">', unsafe_allow_html=True)
# Reset Password Form
if st.session_state.get('show_reset_form', False):
st.markdown('<h1 class="auth-title">Reset Password</h1>', unsafe_allow_html=True)
st.markdown('<p class="auth-subtitle">Enter your username and new password</p>', unsafe_allow_html=True)
with st.form("reset_form"):
username = st.text_input("Username")
new_password = st.text_input("New Password", type="password")
confirm_password = st.text_input("Confirm Password", type="password")
submit = st.form_submit_button("Reset Password")
if submit:
if not username or not new_password or not confirm_password:
st.error("All fields are required.")
elif new_password != confirm_password:
st.error("Passwords do not match.")
elif update_password(username, new_password):
st.success("Password updated successfully!")
st.session_state['show_reset_form'] = False
time.sleep(1)
st.rerun()
else:
st.error("Username not found.")
if st.button("Back to Login"):
st.session_state['show_reset_form'] = False
st.rerun()
# Registration Form
elif st.session_state.get('show_register_form', False):
st.markdown('<h1 class="auth-title">Create Account</h1>', unsafe_allow_html=True)
st.markdown('<p class="auth-subtitle">Sign up to get started</p>', unsafe_allow_html=True)
with st.form("register_form"):
new_username = st.text_input("Username")
new_password = st.text_input("Password", type="password")
submit_button = st.form_submit_button("Create Account")
if submit_button:
if new_username and new_password:
if add_user(new_username, new_password):
st.success("Account created successfully!")
st.session_state['show_register_form'] = False
time.sleep(1)
st.rerun()
else:
st.error("Username already exists.")
else:
st.error("Username and password are required.")
st.markdown('<div class="divider"><span>OR</span></div>', unsafe_allow_html=True)
if st.button("Back to Login"):
st.session_state['show_register_form'] = False
st.rerun()
# Login Form (default)
else:
st.markdown('<h1 class="auth-title">Welcome Back</h1>', unsafe_allow_html=True)
st.markdown('<p class="auth-subtitle">Sign in to your account</p>', unsafe_allow_html=True)
with st.form("login_form"):
username = st.text_input("Username")
password = st.text_input("Password", type="password")
cols = st.columns([1, 1])
submit_button = cols[0].form_submit_button("Sign In")
forgot_password = cols[1].form_submit_button("Forgot Password?")
if submit_button:
if login(username, password):
st.success("Logged in successfully!")
time.sleep(1)
st.rerun()
elif forgot_password:
st.session_state['show_reset_form'] = True
st.rerun()
st.markdown('<div class="divider"><span>OR</span></div>', unsafe_allow_html=True)
if st.button("Create New Account"):
st.session_state['show_register_form'] = True
st.rerun()
st.markdown('</div>', unsafe_allow_html=True)
# Update the footer section (replace the existing footer with this)
st.markdown("""
<div style="background: linear-gradient(to right, #1e293b, #334155); color: white; padding: 40px 0; margin-top: 40px;">
<div style="max-width: 1200px; margin: auto; padding: 0 20px;">
<div style="display: flex; flex-wrap: wrap; justify-content: space-between; gap: 40px;">
<!-- About Section -->
<div style="flex: 1; min-width: 250px;">
<h3 style="color: #60a5fa; font-size: 1.5em; margin-bottom: 20px;">About Our Platform</h3>
<p style="color: #e2e8f0; line-height: 1.6; margin-bottom: 20px;">
Our AI-powered platform revolutionizes crop disease detection and management.
We combine cutting-edge technology with agricultural expertise to protect your crops
and maximize your yield.
</p>
</div>
<div style="flex: 1; min-width: 250px;">
<h3 style="color: #60a5fa; font-size: 1.5em; margin-bottom: 20px;">Key Features</h3>
<ul style="list-style: none; padding: 0; color: #e2e8f0;">
<li style="margin-bottom: 10px; display: flex; align-items: center;">
<span style="color: #60a5fa; margin-right: 10px;">✓</span> Real-time Disease Detection
</li>
<li style="margin-bottom: 10px; display: flex; align-items: center;">
<span style="color: #60a5fa; margin-right: 10px;">✓</span> Multi-language Support
</li>
<li style="margin-bottom: 10px; display: flex; align-items: center;">
<span style="color: #60a5fa; margin-right: 10px;">✓</span> Expert Analysis Reports
</li>
<li style="margin-bottom: 10px; display: flex; align-items: center;">
<span style="color: #60a5fa; margin-right: 10px;">✓</span> Treatment Recommendations
</li>
</ul>
</div>
<div style="flex: 1; min-width: 250px;">
<h3 style="color: #60a5fa; font-size: 1.5em; margin-bottom: 20px;">Contact Us</h3>
<p style="color: #e2e8f0; line-height: 1.6; margin-bottom: 10px;">
<span style="color: #60a5fa;">Email:</span> support@crophealth.ai
</p>
<p style="color: #e2e8f0; line-height: 1.6; margin-bottom: 20px;">
<span style="color: #60a5fa;">Phone:</span> +1 (234) 567-8900
</p>
<div style="display: flex; gap: 15px; margin-top: 20px;">
<a href="#" style="color: #60a5fa; text-decoration: none; font-size: 1.2em;">
<span>📱</span>
</a>
<a href="#" style="color: #60a5fa; text-decoration: none; font-size: 1.2em;">
<span>💬</span>
</a>
<a href="#" style="color: #60a5fa; text-decoration: none; font-size: 1.2em;">
<span>📨</span>
</a>
</div>
</div>
</div>
<div style="border-top: 1px solid #4b5563; margin-top: 40px; padding-top: 20px; text-align: center;">
<p style="color: #e2e8f0; font-size: 0.9em;">
© 2025 Crop Disease Detection System. All rights reserved.
</p>
<div style="margin-top: 10px;">
<a href="#" style="color: #e2e8f0; text-decoration: none; margin: 0 10px; font-size: 0.9em;">Privacy Policy</a>
<a href="#" style="color: #e2e8f0; text-decoration: none; margin: 0 10px; font-size: 0.9em;">Terms of Service</a>
<a href="#" style="color: #e2e8f0; text-decoration: none; margin: 0 10px; font-size: 0.9em;">FAQ</a>
</div>
</div>
</div>
</div>
""", unsafe_allow_html=True)
st.stop()
# Update database schema to include comments
def setup_feedback_db():
conn = sqlite3.connect('customer_feedback.db')
c = conn.cursor()
c.execute('''CREATE TABLE IF NOT EXISTS customer_feedback
(id INTEGER PRIMARY KEY AUTOINCREMENT,
question TEXT,
response TEXT,
feedback_type TEXT,
comment_type TEXT,
custom_comment TEXT,
timestamp DATETIME DEFAULT CURRENT_TIMESTAMP)''')
conn.commit()
return conn, c
def save_feedback(question, response, feedback_type, comment_type=None, custom_comment=None):
conn, c = setup_feedback_db()
try:
c.execute("""INSERT INTO customer_feedback
(question, response, feedback_type, comment_type, custom_comment)
VALUES (?, ?, ?, ?, ?)""",
(question, response, feedback_type, comment_type, custom_comment))
conn.commit()
return True
except Exception as e:
st.error(f"Error saving feedback: {e}")
return False
finally:
conn.close()
# Update the conversation display section
def display_feedback_buttons(file_id, index, question, response):
# Suggested comments
SUGGESTED_COMMENTS = [
"Inaccurate information",
"Unclear explanation",
"Missing details",
"Not relevant to question",
"Technical error",
"Other"
]
# Initialize session state for feedback if it doesn't exist
if f"feedback_{file_id}_{index}" not in st.session_state:
st.session_state[f"feedback_{file_id}_{index}"] = {
"feedback_type": None, # Stores "👍" or "👎"
"comment": None, # Stores the user's comment
"submitted": False # Tracks whether feedback has been submitted
}
col1, col2 = st.columns([1, 4])
with col1:
if st.button("👍", key=f"helpful_{file_id}_{index}"):
# Save positive feedback immediately
save_feedback(question, response, "👍")
st.success("Feedback saved!")
# Update session state to indicate feedback has been submitted
st.session_state[f"feedback_{file_id}_{index}"]["submitted"] = True
return
with col2:
if st.button("👎", key=f"not_helpful_{file_id}_{index}"):
# Store the feedback type in session state
st.session_state[f"feedback_{file_id}_{index}"]["feedback_type"] = "👎"
# Check if feedback_type is "👎" before showing the comment input field
if st.session_state[f"feedback_{file_id}_{index}"].get("feedback_type") == "👎":
# Display suggested comments in a dropdown menu
selected_comment = st.selectbox(
"What was the issue?",
options=SUGGESTED_COMMENTS,
key=f"suggested_comment_{file_id}_{index}"
)
# If the user selects "Other", allow them to provide a custom comment
custom_comment = None
if selected_comment == "Other":
custom_comment = st.text_area(
"Please describe the issue:",
key=f"custom_comment_{file_id}_{index}"
)
# Submit Feedback button
if st.button("Submit Feedback", key=f"submit_{file_id}_{index}"):
# Save feedback to the database
save_feedback(
question,
response,
st.session_state[f"feedback_{file_id}_{index}"]["feedback_type"],
custom_comment if selected_comment == "Other" else selected_comment
)
st.success("Thank you for your feedback!")
# Update session state to indicate feedback has been submitted
st.session_state[f"feedback_{file_id}_{index}"]["submitted"] = True
return
from qdrant_client import QdrantClient
from qdrant_client.http.models import Distance, VectorParams
# Initialize Qdrant client (local instance)
client = QdrantClient(
url="https://f8a5b65d-191d-4a67-8536-ffd96c2f49c6.us-east4-0.gcp.cloud.qdrant.io:6333",
api_key=os.environ['QDRANT_API_KEY']
)
# Collection name
collection_name = "crop_disease_embeddings"
# Check if the collection already exists
existing_collections = client.get_collections()
collection_names = [col.name for col in existing_collections.collections]
if collection_name not in collection_names:
# Create the collection if it doesn't exist
client.recreate_collection(
collection_name=collection_name,
vectors_config=VectorParams(size=384, distance=Distance.COSINE)
)
st.write(f"Created new collection: {collection_name}")
else:
st.write(f"Collection {collection_name} already exists. Skipping creation.")
# Load a pre-trained model (e.g., 'all-MiniLM-L6-v2')
embedding_model = SentenceTransformer('all-MiniLM-L6-v2')
# Read the text file
with open("./docs/disease_docs.txt", "r") as file:
text = file.read()
# Split the text into paragraphs (assuming paragraphs are separated by double newlines)
#paragraphs = text.split("\n\n")
# Alternatively, split into sentences (using a simple approach)
import re
sentences = re.split(r'(?<=[.!?]) +', text)
# Create documents from paragraphs or sentences
documents = []
for idx, chunk in enumerate(sentences): # or sentences
documents.append({
"id": idx + 1, # Unique ID for each chunk
"text": chunk.strip(), # Remove leading/trailing whitespace
"metadata": {
"source": "docs/disease_docs.txt", # Add any relevant metadata
"chunk_number": idx + 1
}
})
# Generate embeddings and index them
for doc in documents:
embedding = embedding_model.encode(doc["text"])
client.upsert(
collection_name=collection_name,
points=[
{
"id": doc["id"],
"vector": embedding.tolist(),
"payload": {
"text": doc["text"], # Include the text in the payload
"source": doc["metadata"]["source"],
"chunk_number": doc["metadata"]["chunk_number"]
}
}
]
)
def retrieve_relevant_documents(query, detected_diseases, top_k=3):
"""
Retrieve relevant documents based on the user's query and detected diseases.
"""
# Combine the user's query with the detected diseases
combined_query = f"{query} {', '.join(detected_diseases)}"
# Generate embeddings for the combined query
query_embedding = embedding_model.encode(combined_query)
# Search the Qdrant database
search_result = client.search(
collection_name=collection_name,
query_vector=query_embedding.tolist(),
limit=top_k
)
# Log the retrieved chunks
#st.write("Retrieved relevant chunks:")
#for chunk in search_result:
# st.write(f"- {chunk.payload['text']}")
return [hit.payload for hit in search_result]
def get_reference_answer(disease_name):
"""
Retrieve the reference answer (cause, symptoms, and treatment) for a specific disease from the SQLite database.
"""
conn = sqlite3.connect('./db/disease_knowledge_base.db') # Replace with your database path
c = conn.cursor()
# Query the database for cause, symptoms, and treatment
c.execute("SELECT cause, symptoms, treatment FROM diseases WHERE name = ?", (disease_name,))
result = c.fetchone()
conn.close()
if result:
# Format the result into a structured response
cause, symptoms, treatment = result
reference_answer = (
f"Cause: {cause}\n"
f"Symptoms: {symptoms}\n"
f"Treatment: {treatment}"
)
return reference_answer # Return the formatted reference answer
else:
return None # Return None if no matching disease is found
def filter_relevant_chunks(chunks, detected_diseases):
"""
Filter retrieved chunks to include only those relevant to the detected diseases.
Args:
chunks (list): List of retrieved chunks (each chunk is a dictionary with a "text" key).
detected_diseases (list): List of detected disease names.
Returns:
list: Filtered list of chunks relevant to the detected diseases.
"""
filtered_chunks = []
for chunk in chunks:
# Check if the chunk text contains any of the detected diseases
if any(disease.lower() in chunk["text"].lower() for disease in detected_diseases):
filtered_chunks.append(chunk)
return filtered_chunks
async def generate_rag_response(query, conversation_history=None, reference_answer=None):
"""
Generate a response using RAG and evaluate it using RAGAS metrics.
"""
# Retrieve relevant chunks
relevant_chunks = retrieve_relevant_documents(query, detected_classes)
# Filter the retrieved chunks to include only those relevant to the detected diseases
filtered_chunks = filter_relevant_chunks(relevant_chunks, detected_classes)
# Build context from filtered chunks
context = "\n".join([chunk["text"] for chunk in filtered_chunks])
# Generate response using Ollama
response = await generate_groq_response(
f"Context: {context}\n\nQuestion: {query}",
model_name=selected_model,
conversation_history=conversation_history
)
# Evaluate using the new local implementation
ragas_result = evaluate_ragas(query, response, context, reference_answer)
# Evaluate the generation-only system
generation_only_result = evaluate_generation_only(query, reference_answer)
print("Full RAG System Results:")
print(f"Answer Relevancy: {ragas_result['answer_relevancy']:.2f}")
print(f"Faithfulness: {ragas_result['faithfulness']:.2f}")
print(f"Answer Correctness: {ragas_result['answer_correctness']:.2f}")
print(f"Context Precision: {ragas_result['context_precision']:.2f}")
print(f"Context Recall: {ragas_result['context_recall']:.2f}")
print("Generation-Only System Results:")
print(f"Answer Relevancy: {generation_only_result['answer_relevancy']:.2f}")
print(f"Faithfulness: {generation_only_result['faithfulness']:.2f}")
if generation_only_result['answer_correctness'] is not None:
print(f"Answer Correctness: {generation_only_result['answer_correctness']:.2f}")
else:
print("Answer Correctness: N/A (No reference answer provided)")
# Display metrics
st.markdown(format_evaluation_results(ragas_result))
return response, filtered_chunks, ragas_result
def generate_answer_without_retrieval(query, model_name="llama2"):
"""
Generate an answer using only the LLM (no retrieval).
"""
response = asyncio.run(generate_groq_response(query, model_name=model_name))
return response
def evaluate_generation_only(query, reference_answer=None):
"""
Evaluate the generation-only system using RAGAS metrics.
"""
# Generate answer without retrieval
response = generate_answer_without_retrieval(query)
# Evaluate metrics that don't require context
evaluator = LocalMetricsEvaluator()
answer_relevancy = evaluator.evaluate_answer_relevancy(query, response)
faithfulness = 1.0 # No context to compare, assume perfect faithfulness
# Evaluate answer correctness only if reference_answer is provided
answer_correctness = None
if reference_answer:
answer_correctness = evaluator.evaluate_answer_correctness(response, reference_answer)
return {
"answer_relevancy": answer_relevancy,
"faithfulness": faithfulness,
"answer_correctness": answer_correctness
}
class LocalMetricsEvaluator:
def __init__(self, embedding_model: str = "sentence-transformers/all-MiniLM-L6-v2"):
self.embeddings = SentenceTransformer(embedding_model)
def calculate_semantic_similarity(self, text1: str, text2: str) -> float:
# Calculate embeddings
emb1 = self.embeddings.encode(text1)
emb2 = self.embeddings.encode(text2)
# Calculate cosine similarity
similarity = np.dot(emb1, emb2) / (np.linalg.norm(emb1) * np.linalg.norm(emb2))
return float(similarity)
def evaluate_answer_relevancy(self, question: str, answer: str) -> float:
return self.calculate_semantic_similarity(question, answer)
def evaluate_faithfulness(self, answer: str, context: str) -> float:
return self.calculate_semantic_similarity(answer, context)
def evaluate_answer_correctness(self, answer: str, reference: str) -> float:
return self.calculate_semantic_similarity(answer, reference)
def evaluate_context_precision(self, question: str, context: str) -> float:
return self.calculate_semantic_similarity(question, context)
def evaluate_context_recall(self, question: str, retrieved_context: str, reference: str) -> float:
"""
Evaluate context recall by comparing the retrieved context to the ground truth context.
Args:
question (str): The user's question.
retrieved_context (str): The context retrieved by the RAG system.
ground_truth_context (str): The ground truth context (relevant information).
Returns:
float: A score between 0 and 1, representing how well the retrieved context covers the ground truth context.
"""
# Calculate embeddings for the retrieved context and ground truth context
retrieved_embedding = self.embeddings.encode(retrieved_context)
ground_truth_embedding = self.embeddings.encode(reference)
# Calculate cosine similarity between the retrieved context and ground truth context
recall_score = np.dot(retrieved_embedding, ground_truth_embedding) / (
np.linalg.norm(retrieved_embedding) * np.linalg.norm(ground_truth_embedding)
)
return float(recall_score)
def evaluate_ragas(query: str, response: str, context: str, reference_answer: str = None):
"""
Evaluate the RAG system using local embeddings instead of OpenAI.
Args:
query (str): The user's question
response (str): The generated response
context (str): The context used to generate the response
reference_answer (str, optional): Ground truth answer
Returns:
dict: Dictionary containing evaluation metrics
"""
# Initialize evaluator
evaluator = LocalMetricsEvaluator()
# If no reference answer is provided, use the response
if reference_answer is None:
reference_answer = response
# Calculate metrics
metrics = {
"answer_relevancy": evaluator.evaluate_answer_relevancy(query, response),
"faithfulness": evaluator.evaluate_faithfulness(response, context),
"answer_correctness": evaluator.evaluate_answer_correctness(response, reference_answer),
"context_precision": evaluator.evaluate_context_precision(query, context),
"context_recall": evaluator.evaluate_context_recall(query, context, reference_answer)
}
return metrics
def format_evaluation_results(metrics: Dict[str, float]) -> str:
"""Format the evaluation results for display"""
return "\n".join([
f"📊 RAGAS Evaluation Results:",
f"• Answer Relevancy: {metrics['answer_relevancy']:.3f}",
f"• Faithfulness: {metrics['faithfulness']:.3f}",
f"• Answer Correctness: {metrics['answer_correctness']:.3f}",
f"• Context Precision: {metrics['context_precision']:.3f}",
f"• Context Recall: {metrics['context_recall']:.3f}"
])
# Model configuration
SUPPORTED_MODELS = {
"deepseek-r1-distill-llama-70b": {
"name": "deepseek-r1-distill-llama-70b",
"system_prompt": "You are a helpful plant pathology expert assistant.",
"supports_vision": False
},
"mistral-saba-24b": {
"name": "mixtral-8x7b-32768",
"system_prompt": "You are a helpful plant pathology expert assistant.",
"supports_vision": False
},
"qwen/qwen3-32b": {
"name": "qwen/qwen3-32b",
"system_prompt": "You are a helpful plant pathology expert assistant.",
"supports_vision": False
},
"gemma2-9b-it": {
"name": "gemma2-9b-it",
"system_prompt": "You are a helpful plant pathology expert assistant.",
"supports_vision": False
},
"llama-3.1-8b-instant": {
"name": "llama-3.1-8b-instant",
"system_prompt": "You are a helpful plant pathology expert assistant.",
"supports_vision": False
},
"llama-3.3-70b-versatile": {
"name": "llama-3.3-70b-versatile",
"system_prompt": "You are a helpful plant pathology expert assistant.",
"supports_vision": False
},
"llama3-70b-8192": {
"name": "llama3-70b-8192",
"system_prompt": "You are a helpful plant pathology expert assistant.",
"supports_vision": False
}
}
# Initialize session state for conversation history if it doesn't exist
if 'conversation_history' not in st.session_state:
st.session_state.conversation_history = {}
# Load YOLOv8 model
yolo_model = YOLO("./model/plantdoc_model_yolov8.pt")
def preprocess_image(image, target_size=(224, 224)):
"""
Preprocess the image for vision-capable models.
"""
image = Image.fromarray(image)
image = image.resize(target_size)
return image
def text_to_speech(text, voice="af_heart", language="en"):
"""Convert text to speech using Kokoro TTS with local voice files."""
try:
# Initialize Kokoro pipeline
pipeline = KPipeline(lang_code="a") # 'a' for American English, 'b' for British
# Ensure selected voice exists
voice_path = os.path.join(VOICES_DIR, f"{voice}.pt")
print(f"Loading voice file: {voice_path}") # Debugging step
if not os.path.exists(voice_path):
raise FileNotFoundError(f"Voice file '{voice_path}' not found. Please check your Kokoro directory.")
# Generate speech
generator = pipeline(text, voice=voice, speed=1, split_pattern=r"\n+")
audio_data = []
for _, _, audio in generator:
audio_data.extend(audio)
audio_array = np.array(audio_data, dtype=np.float32)
# Save to a temporary file
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
sf.write(temp_audio.name, audio_array, 24000)
# Read the audio file
with open(temp_audio.name, 'rb') as audio_file:
audio_bytes = audio_file.read()
os.unlink(temp_audio.name) # Clean up temp file
return audio_bytes
except FileNotFoundError as e:
st.error(f"Error: {str(e)}")
return None
except Exception as e:
st.error(f"Error generating speech: {str(e)}")
return None
async def generate_rag_response_general(query, conversation_history=None):
"""
Generate a response using RAG for general questions (no specific detected diseases)
"""
# Retrieve relevant chunks based on the query
relevant_chunks = retrieve_relevant_documents(query, [], top_k=5) # Empty disease list for general queries
# Build context from retrieved chunks
context = "\n".join([chunk["text"] for chunk in relevant_chunks])
# Create a more general prompt for consultation
consultation_prompt = f"""As an expert plant pathologist and agricultural consultant, please provide a comprehensive answer to the following question about crop diseases and plant health.
Context from knowledge base:
{context}
Question: {query}
Please provide a detailed, practical response that includes:
1. Direct answer to the question
2. Relevant scientific background
3. Practical recommendations
4. Prevention strategies (if applicable)
5. When to seek professional help (if applicable)
Make your response accessible to farmers and agricultural practitioners while maintaining scientific accuracy."""
# Generate response
selected_model_name = SUPPORTED_MODELS[st.session_state.get('selected_model', 'llama-3.1-8b-instant')]["name"]
response = await generate_groq_response(
consultation_prompt,
model_name=selected_model_name,
conversation_history=conversation_history
)
# Evaluate using local metrics (simplified for general consultation)
evaluator = LocalMetricsEvaluator()
ragas_result = {
"answer_relevancy": evaluator.evaluate_answer_relevancy(query, response),
"faithfulness": evaluator.evaluate_faithfulness(response, context),
"answer_correctness": 0.8, # Placeholder since we don't have ground truth for general questions
"context_precision": evaluator.evaluate_context_precision(query, context),
"context_recall": 0.8 # Placeholder
}
return response, relevant_chunks, ragas_result
async def generate_groq_response(prompt, model_name="mixtral-8x7b-32768", conversation_history=None):
try:
# Build the messages array
messages = [
{"role": "system", "content": "You are a helpful plant pathology expert assistant."}
]
# Add conversation history
if conversation_history:
for entry in conversation_history:
if len(entry) >= 2: # Handle tuples with 2 or 3 values
question, response = entry[:2]
messages.extend([
{"role": "user", "content": question},
{"role": "assistant", "content": response}
])
# Add the current prompt
messages.append({"role": "user", "content": prompt})
# Generate response using Groq
response = groq_client.chat.completions.create(
model=model_name,
messages=messages
)
return response.choices[0].message.content
except Exception as e:
return f"Error connecting to Groq: {str(e)}"
def generate_improved_description(detected_classes, class_names, user_text, image_details=None, conversation_history=None):
"""
Generate a more detailed and contextual description using Ollama
"""
detected_objects = [class_names[cls] for cls in detected_classes]
# Create base context about detected diseases
disease_context = f"Detected diseases: {', '.join(detected_objects)}"
# Different prompt structure for initial vs. follow-up questions
if not conversation_history:
base_prompt = f"""As an expert plant pathologist, analyze the following crop diseases detected in the provided image: {', '.join(detected_objects)}.
For each detected disease, provide a structured analysis following this format:
1. Disease Name: [Name]
- Pathogen: [Causative organism]
- Severity Level: [Based on visual symptoms]
- Key Symptoms:
* [Symptom 1]
* [Symptom 2]
- Economic Impact:
* [Brief description of potential crop losses]
- Treatment Options:
* Immediate actions: [Short-term solutions]
* Long-term management: [Preventive measures]
- Environmental Conditions:
* Favorable conditions for disease development
* Risk factors
2. Recommendations:
- Immediate Steps:
* [Action items for immediate control]
- Prevention Strategy:
* [Long-term prevention measures]
- Monitoring Protocol:
* [What to watch for]
Initial Question/Context: {user_text if user_text else "Provide a general analysis"}
"""
else:
base_prompt = f"""Context: {disease_context}
Previous conversation context has been provided above. Please address the following follow-up question while maintaining consistency with previous responses:
{user_text}
Provide a detailed response that builds upon the previous context and specifically addresses this question."""
# Get the selected model from session state or default to llama2
selected_model = st.session_state.get('selected_model', 'llama2')
return asyncio.run(generate_groq_response(
base_prompt,
model_name=selected_model,
conversation_history=conversation_history,
#image_data=image_details.get("image_data") if image_details else None
))
def inference(image):
"""
Enhanced inference function with confidence scores and bounding box information
"""
results = yolo_model(image, conf=0.4)
infer = np.zeros(image.shape, dtype=np.uint8)
classes = dict()
names_infer = []
confidence_scores = []
bounding_boxes = []
for r in results:
infer = r.plot()
classes = r.names
names_infer = r.boxes.cls.tolist()
confidence_scores = r.boxes.conf.tolist()
bounding_boxes = r.boxes.xyxy.tolist()
return infer, names_infer, classes, confidence_scores, bounding_boxes
# Streamlit application
st.sidebar.markdown("---")
st.sidebar.header("🔬 Research Tools")
# Page selection
page_selection = st.sidebar.radio(
"Navigate to:",
["🏠 Main App", "🔬 Research Dashboard"],
index=0
)
if page_selection == "🔬 Research Dashboard":
render_research_page()
else:
# Your existing main app code
st.title("Interactive Crop Disease Detection and Analysis🌾🌿🥬☘️")
st.write(f"Welcome, {st.session_state['username']}!😊")
# Logout button
if st.button("Logout"):
logout()
st.rerun()
# Add sidebar for configuration
with st.sidebar:
st.header("Settings")
selected_model = st.selectbox(
"Select LLM Model",
list(SUPPORTED_MODELS.keys()),
help="Choose the Ollama model to use for analysis"
)
# Store the selected model in session state
st.session_state['selected_model'] = selected_model
if SUPPORTED_MODELS[selected_model]["supports_vision"]:
st.info("This model supports vision capabilities and can analyze images directly.")
confidence_threshold = st.slider("Detection Confidence Threshold", 0.0, 1.0, 0.4)
show_confidence = st.checkbox("Show Confidence Scores", value=True)
show_bbox = st.checkbox("Show Bounding Boxes", value=True)
# TTS Settings
# Path to your Kokoro repository
KOKORO_DIR = "./Kokoro-82M"
VOICES_DIR = os.path.join(KOKORO_DIR, "voices")
# Automatically list available voices
available_voices = [f.replace(".pt", "") for f in os.listdir(VOICES_DIR) if f.endswith(".pt")]
# Ensure there are voices available
if not available_voices:
available_voices = ["af_heart"] # Default fallback voice if directory is empty
# Streamlit voice selection dropdown
selected_voice = st.sidebar.selectbox("Choose a Voice", available_voices, index=0)
# Add option to clear conversation history
if st.button("Clear All Conversations"):
st.session_state.conversation_history = {}
st.success("Conversation history cleared!")
# Initialize translator
translator = Translator()
# Language selection
language = st.selectbox(
"Select Language",
options=['en', 'es', 'fr', 'de', 'ak', 'gaa', 'ee'], # Add more languages as needed
format_func=lambda x: {
'en': 'English',
'es': 'Spanish',
'fr': 'French',
'de': 'German',
'ak': 'Twi',
'gaa': 'Ga',
'ee': 'Ewe'
}[x],
help="Select your preferred language"
)
tab1, tab2 = st.tabs(["🖼️ Image Analysis", "💬 General Consultation"])
with tab1:
st.header("Image-Based Disease Detection")
st.write("Upload images of your crops to detect diseases and get specific analysis.")
# Main content - Image upload and analysis
uploaded_files = st.file_uploader("Upload images for disease detection", type=["jpg", "jpeg", "png"], accept_multiple_files=True)
if uploaded_files:
for uploaded_file in uploaded_files:
file_id = uploaded_file.name
# Initialize conversation history for this image if it doesn't exist
if file_id not in st.session_state.conversation_history:
st.session_state.conversation_history[file_id] = []
st.subheader(f"Analysis for {file_id}")
# Create columns for side-by-side display
col1, col2 = st.columns(2)
# Process image
file_bytes = np.asarray(bytearray(uploaded_file.read()), dtype=np.uint8)
image = cv2.imdecode(file_bytes, 1)
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# Display original image
with col1:
st.subheader("Original Image")
st.image(image)
# Process and display results
with st.spinner("Processing image..."):
infer_image, classes_in_image, classes_in_dataset, confidences, boxes = inference(image)
with col2:
st.subheader("Detected Diseases")
st.image(infer_image)
# Display detection details
if show_confidence:
st.subheader("Detection Details")
for cls, conf in zip(classes_in_image, confidences):
st.write(f"- {classes_in_dataset[cls]}: {conf:.2%} confidence")
# Display conversation history for this image
if st.session_state.conversation_history[file_id]:
st.subheader("Conversation History")
for i, entry in enumerate(st.session_state.conversation_history[file_id]):
question, response = entry[:2]
with st.expander(f"Q{i+1}: {question[:50]}...", expanded=False):
st.write("**Question:**", question)
st.write("**Response:**", response)
# Display feedback buttons and handle comment collection
display_feedback_buttons(file_id, i, question, response)
# Audio playback option
if st.button("🔊 Listen", key=f"listen_history_{file_id}_{i}"):
with st.spinner("Generating audio..."):
audio_bytes = text_to_speech(response, voice=selected_voice)
if audio_bytes:
st.audio(audio_bytes, format="audio/wav")
# User input for questions about the detected diseases
st.subheader("Ask Questions About Detected Diseases")
user_text = st.text_area(
"Enter your question about the detected diseases:",
placeholder="Example: What are the best treatment options for these diseases? What preventive measures should I take?",
key=f"question_{file_id}"
)
if st.button("Get Analysis", key=f"analyze_{file_id}"):
with st.spinner(f"Generating analysis using {selected_model}..."):
# Translate user input
translated_input = asyncio.run(translator.translate(user_text, dest='en')).text
st.write(f"Translated Input (to English): {translated_input}")
# Extract detected disease names
detected_classes = [classes_in_dataset[cls] for cls in classes_in_image]
# Fetch reference answers for detected diseases
reference_answers = []
for disease_name in detected_classes:
reference_answer = get_reference_answer(disease_name)
if reference_answer:
reference_answers.append(reference_answer)
# Combine reference answers into a single string
reference_answer = "\n".join(reference_answers) if reference_answers else None
# Generate response with RAG
response, relevant_chunks, ragas_result = asyncio.run(generate_rag_response(
translated_input,
st.session_state.conversation_history[file_id],
reference_answer # Pass the reference answer for evaluation
))
print("Response:", response)
if response is None:
st.error("Failed to generate a response. Please try again.")
response = "No response generated."
# Move the translate function call here
if response:
try:
translated_response = asyncio.run(translator.translate(response, dest=language)).text
except Exception as e:
st.error(f"Translation failed: {e}")
translated_response = response # Fallback to the original response
else:
translated_response = response
st.session_state.conversation_history[file_id].append((user_text, translated_response, None))
# Display the response and evaluation metrics
#st.markdown("### Relevant Information")
#for chunk in relevant_chunks:
# st.write(f"- **Chunk {chunk['chunk_number']}**: {chunk['text']}")
st.markdown(response)
# Add audio playback option for the latest response
col1, col2 = st.columns([1, 4])
with col1:
if st.button("🔊 Listen", key=f"listen_latest_{file_id}"):
with st.spinner("Generating audio..."):
audio_bytes = text_to_speech(response, language)
if audio_bytes:
st.audio(audio_bytes, format='audio/mp3')
with tab2:
st.header("General Disease Consultation")
st.write("Ask questions about crop diseases without uploading images. Get expert advice on plant pathology topics.")
# Initialize general consultation history
if 'general_consultation' not in st.session_state.conversation_history:
st.session_state.conversation_history['general_consultation'] = []
# Disease selection helper
st.subheader("🎯 Quick Disease Lookup")
col1, col2 = st.columns([2, 1])
with col1:
# Get list of diseases from database for quick selection
try:
conn = sqlite3.connect('./db/disease_knowledge_base.db')
c = conn.cursor()
c.execute("SELECT DISTINCT name FROM diseases ORDER BY name")
available_diseases = [row[0] for row in c.fetchall()]
conn.close()
except:
available_diseases = ["Corn Leaf Blight", "Apple Scab", "Tomato Late Blight", "Wheat Rust"]
selected_disease = st.selectbox(
"Select a specific disease for quick information:",
[""] + available_diseases,
help="Choose a disease to get instant information about it"
)
with col2:
if selected_disease and st.button("Get Disease Info", key="quick_disease_info"):
with st.spinner("Retrieving disease information..."):
quick_query = f"Tell me about {selected_disease} - its causes, symptoms, and treatment options."
# Generate response using RAG
response, relevant_chunks, ragas_result = asyncio.run(generate_rag_response_general(
quick_query,
st.session_state.conversation_history['general_consultation']
))
# Translate if needed
if language != 'en':
try:
translated_response = translator.translate(response, dest=language).text
except:
translated_response = response
else:
translated_response = response
# Add to conversation history
st.session_state.conversation_history['general_consultation'].append((quick_query, translated_response))
st.markdown("### Disease Information")
st.markdown(translated_response)
# Audio option
if st.button("🔊 Listen to Response", key="listen_quick_disease"):
with st.spinner("Generating audio..."):
audio_bytes = text_to_speech(translated_response, voice=selected_voice)
if audio_bytes:
st.audio(audio_bytes, format="audio/wav")
# General question input
st.subheader("💡 Ask Any Question About Crop Diseases")
# Provide example questions
example_questions = [
"What are the most common fungal diseases in tomatoes?",
"How can I prevent wheat rust in my field?",
"What's the difference between bacterial and viral plant diseases?",
"Which organic treatments work best for aphid control?",
"What are the early signs of nutrient deficiency in corn?",
"How do weather conditions affect plant disease development?",
]
with st.expander("💡 Example Questions", expanded=False):
for i, example in enumerate(example_questions):
if st.button(example, key=f"example_{i}"):
st.session_state[f"general_question_input"] = example
general_question = st.text_area(
"Enter your question about crop diseases, plant pathology, or agricultural practices:",
placeholder="Example: What are the most effective organic methods to control powdery mildew in grapes?",
key="general_question_input",
height=100
)
# Topic categories for better organization
st.subheader("🏷️ Question Categories")
col1, col2, col3 = st.columns(3)
# Define callback functions
def set_treatment_question():
st.session_state.general_question_input = "What are the most effective treatment options for fungal plant diseases?"
def set_identification_question():
st.session_state.general_question_input = "How can I identify different types of plant diseases based on symptoms?"
def set_prevention_question():
st.session_state.general_question_input = "What preventive measures can I take to protect my crops from diseases?"
# Then modify your button calls:
with col1:
st.button("🦠 Disease Identification", key="cat_identification", on_click=set_identification_question)
with col2:
st.button("💊 Treatment Options", key="cat_treatment", on_click=set_treatment_question)
with col3:
st.button("🛡️ Prevention Methods", key="cat_prevention", on_click=set_prevention_question)
if st.button("Get Expert Answer", key="general_analyze", type="primary"):
if general_question.strip():
with st.spinner(f"Consulting plant pathology expert using {selected_model}..."):
# Translate user input if needed
if language != 'en':
try:
translated_input = translator.translate(general_question, dest='en').text
st.info(f"Translated to English: {translated_input}")
except:
translated_input = general_question
else:
translated_input = general_question
# Generate response using RAG for general consultation
response, relevant_chunks, ragas_result = asyncio.run(generate_rag_response_general(
translated_input,
st.session_state.conversation_history['general_consultation']
))
if response:
# Translate response back to user's language
if language != 'en':
try:
translated_response = translator.translate(response, dest=language).text
except Exception as e:
st.error(f"Translation failed: {e}")
translated_response = response
else:
translated_response = response
# Add to conversation history
st.session_state.conversation_history['general_consultation'].append((general_question, translated_response))
# Display response
st.markdown("### Expert Response")
st.markdown(translated_response)
# Show relevant sources if available
if relevant_chunks:
with st.expander("📚 Information Sources", expanded=False):
for i, chunk in enumerate(relevant_chunks[:3]): # Show top 3 sources
st.write(f"**Source {i+1}:** {chunk['text'][:200]}...")
# Audio playback option
col1, col2 = st.columns([1, 4])
with col1:
if st.button("🔊 Listen", key="listen_general_latest"):
with st.spinner("Generating audio..."):
audio_bytes = text_to_speech(translated_response, voice=selected_voice)
if audio_bytes:
st.audio(audio_bytes, format="audio/wav")
else:
st.error("Failed to generate a response. Please try again.")
else:
st.warning("Please enter a question before submitting.")
# Display general consultation history
if st.session_state.conversation_history['general_consultation']:
st.subheader("📝 Consultation History")
for i, entry in enumerate(st.session_state.conversation_history['general_consultation']):
question, response = entry[:2]
with st.expander(f"Q{i+1}: {question[:60]}...", expanded=False):
st.write("**Question:**", question)
st.write("**Response:**", response)
# Feedback buttons for general consultation
display_feedback_buttons('general_consultation', i, question, response)
# Audio playback for history
if st.button("🔊 Listen", key=f"listen_general_history_{i}"):
with st.spinner("Generating audio..."):
audio_bytes = text_to_speech(response, voice=selected_voice)
if audio_bytes:
st.audio(audio_bytes, format="audio/wav")
# Export general consultation
if st.session_state.conversation_history['general_consultation']:
if st.button("📄 Export Consultation", key="export_general"):
consultation_text = f"""
# General Crop Disease Consultation Report
## Consultation Information
- Date: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}
- Language: {language}
- Model Used: {selected_model}
## Consultation History
"""
for i, entry in enumerate(st.session_state.conversation_history['general_consultation']):
question, response = entry[:2]
consultation_text += f"\n### Question {i+1}:\n{question}\n\n### Expert Response {i+1}:\n{response}\n\n---\n"
st.download_button(
label="📥 Download Consultation Report",
data=consultation_text,
file_name=f"crop_disease_consultation_{datetime.now().strftime('%Y%m%d_%H%M%S')}.md",
mime="text/markdown",
key="download_general"
)
# Add a footer with clear instructions
st.markdown("""
---
### How to Use
1. Upload one or more images of crops with potential diseases
2. View the detected diseases and their confidence scores
3. Ask questions about the diseases, treatments, or prevention
4. Use the 🔊 Listen button to hear the responses
5. View previous questions and answers in the conversation history
6. Export the entire conversation for future reference
7. Use the sidebar to adjust settings or clear conversation history
""") |